GVE’99

COMPUTER GRAPHICS

AND

VISUALIZATION EDUCATION '99

Charting the Future for Computer Graphics Education

July 3-5, Coimbra, Portugal

EUROGRAPHICS Workshop

co-sponsored by ACM SIGGRAPH

PROCEEDINGS

Editors

José Carlos TeixeiraWerner HansmannMichael B. McGrath

This work is supported by the National Science Foundation, under AWARD 9906124. All opinions, findings, conclusions andrecommendations in any material resulting from this work are those of the principal investigators, and do not necessarily reflectthe views of the National Science Foundation.

GVE’99 – Coimbra - Portugal i

PREFACE

Welcome to GVE’99! Welcome to Coimbra!

Following several Eurographics education workshops held in different Europeancities, the Computer Graphics and Visualization Education '99, under the motto“Charting the Future for Computer Graphics Education”, will be held in Coimbra,Portugal, from 3th to 5th July. Eurographics organized this workshop in cooperationwith ACM SIGGRAPH and with the support of the National Science Foundation.Computer graphics and visualization education specialists will meet to share theirexperiences and their visions and to make projections andrecommendations for the next ten years of computer graphics education.43 position papers were submitted from 15 countries. The excellent position paperswe received promise a very interesting event. People were invited to participate in theworkshop according to their contributions. Based on the contributions and theworkshop goals, the IPC decided that the workshop should be organised so thatparticipants spend significative time working in group. Group discussion will be oneof the most important parts in the workshop, addressing different educational issues.The participants will be free to choose the working groups they are most interested inor they feel they can best contribute to. The main goal of the few formal presentationswe will have is to give broad overviews and start the discussions.Considerable time will be spent discussing future directions for computer graphicsand visualization education. The recommendations and predictions from thisworkshop will be distributed at both SIGGRAPH’99 and Eurographics’99conferences as part of their joint celebration of SIGGRAPH's 30thanniversary and Eurographics' 20th anniversary.Based on the position papers quality, some of them will be selected for revision andlater published in the journal Computers & Graphics.

The workshop co-chairmen would like to thank all the speakers for their valuableeffort in preparing their formal presentations; the participants for their contributions;the IPC members for the effort in promoting and structuring this event; the GVE’99Committee for its continuing and active support during all the preparation phase; andthe workshop organisation committee for its committed and creative effort to makethe GVE’99 a pleasant workshop.

We welcome you and hope you enjoy GVE’99.

Coimbra, July 1999.

José Carlos Teixeira, Werner Hansmann, Michael B. McGrath

GVE’99 – Coimbra - Portugal iii

INTERNATIONAL PROGRAMME COMMITTEE

Ana Luisa Solis, UNAM, Mexico City, Mexico

Anne Mumford, Loughborough University, UK

Beatriz Sousa Santos, University of Aveiro, Portugal

Bianca Falcidieno, CNR, Italy

Charles Wuethrich, Bauhaus University Weimar, Germany

Danail Dochev, IIT-BAS, Bulgaria

Dena Eber, Bowling Green State University, USA

Dieter Fellner, Braunschweig University of Technology, Germany

Gary Bertoline, Purdue University, USA

Gitta Domik, University of Paderborn, Germany

Jack Bresenham, Winthrop University, USA

Jacquelyn Morie, Rhythm & Hues, USA

Jiaoying Shi, Zhejiang University, Hangzhou, China

John Patterson, University of Glasgow, UK

José Encarnação, Technical University of Darmstadt, Germany

Judith R. Brown, The University of Iowa, USA

Ken Brodlie, University of Leeds, UK

Lars Kjelldahl, Royal Institute of Technology, Sweden

Lewis Hitchner, California Polytechnic State Univ., San Luis Obispo, USA

Mário Rui Gomes, IST - Instituto Superior Técnico, Portugal

Michael Joseph Zyda, Naval Postgraduate School, USA

Pere Brunet, Polytechnic University of Catalunya, Spain

Peter Stockinger, MSH, France

Roger Hubbold, Manchester University, UK

Rosalee Wolfe, DePaul University, USA

Scott Owen, Georgia State University, USA

Stavros Christodoulakis, TUC/MUSIC, Greece

Steve Cunningham, California State University Stanislaus, USA

Vaclav Skala, University of West Bohemia, Czech Republic

Yuri Bayakovski, Moscow State University, Russia

GVE’99 – Coimbra - Portugal v

ORGANIZATION

CCG – Computer Graphics Center – Coimbra

WORKSHOP CO-CHAIRMEN

José Carlos TeixeiraDep. Matemática - F.C.T. - Universidade de Coimbra

Werner HansmannFB Informatik/TIS - Universität Hamburg

Michael B. McGrath (ACM SIGGRAPH Contact)Engineering Department - Colorado School of Mines

WORKSHOP ORGANISATION COMMITTEE

Joaquim Madeira (FCTUC)

César Páris

Sónia Branco

Paula Santos (Secretary)

Vítor Varalonga

GVE’99 – Coimbra - Portugal vii

Index

Keynote Speech

Educational Trojan HorsesMichael J. Bailey

1

1 - Computer Graphics Curricula in Computer Science

Presentation

Bringing the Introductory Computer Graphics Course into the 21stCenturyRosalee Wolfe

3

1.1 - Curricula

Position Papers

An Introductory Course on VisualizationB. Sousa Santos

9

Restructuring Computer Graphics and Visualisation Curricula at ISTJ. M. Brisson Lopes, M. R. Gomes, J. C. Bernardo, J. Jorge and J. M.Pereira

17

Adapting Computer Graphics Curricula to Changes in GraphicsTechnologyL. E. Hitchner and H. A. Sowizral

23

Computer Graphics in the Scope of Informatics Engineering educationM. Próspero dos Santos

31

A Course in Topology-based Geometric ModelingP. Lienhardt, L. Fuchs and Y. Bertrand

39

Re-Inventing the Introductory Computer Graphics Course: ProvidingTools for a Wider AudienceS. Cunningham

45

1.2 - Techniques and Tools

Position Papers

A Mathematica Package for CAGD and Computer GraphicsA. Iglesias, F. Gutiérrez and A. Gálvez

51

The Development of a Digital Library to Support the Teaching ofComputer Graphics and VisualizationG. Scott Owen, Raj Sunderraman and Yanqing Zhang

59

viii GVE’99 – Coimbra - Portugal

Learning 3D Visualisation with MALUDAJ. A. Madeiras Pereira and M. R. Gomes

65

Teaching the Graphics Processing Pipeline: Cosmetic and GeometricAttribute ImplicationsJ. Bresenham

71

Strengthening Visual Skills by Recognising Rendering AlgorithmsR. Wolfe

79

MAVERIK: A Virtual Reality System for Research and TeachingT. Howard, R. Hubbold and A. Murta

85

2 - Computer Graphics Curricula in the Arts

Presentations

Computer Graphics Curricula in the Visual ArtsD. E. Eber

93

An Interdisciplinary Approach to Teaching Computer Animation toArtists and Computer ScientistsD. S. Ebert and D. Bailey

99

Position Papers

Combining Art and Mathematics in Computer Graphics EducationM. Ollila

105

Computers in Art and Design Education - Past, Present and FutureT. Longson

111

3 - Online Education

Presentation

Online EducationAnne Mumford

115

3.1 - Architecture

Position Paper

On the Functionality of Distributed Multimedia-System Architecturefor Educational ApplicationsD. Dochev, R. Pavlov and R. Yoshinov

117

GVE’99 – Coimbra - Portugal ix

3.2 - Collaboration

Position Papers

Three Collaborative Models for Computer Graphics EducationC. Case

123

Enabling Educational Collaboration - A New Shared RealityJ. R. Brown

131

3.3 - Teaching

Position Papers

Supporting Face to Face TeachingA. Mumford

135

Web Based Multi-User Interactive Graphics Worlds for EducationG. Scott Owen

137

4 - Computer Graphics as a Tool for Teaching & Learning

Presentation

Using Web-Based Computer Graphics to Teach SurgeryK. Brodlie, N. El-Khalili and Ying Li

141

Position Papers

Visualisation in teaching-learning mathematics: the role of the computerG. Chiappini and R. M. Bottino

147

Virtual Environment of Water Molecules for Learning and TeachingScienceJ. F. Trindade, C. Fiolhais, V. Gil and J. C. Teixeira

153

Evaluation of Driving Education Methods in a Driving SimulatorJ. M. Leitão, A. Moreira, J. A. Santos, A. A. Sousa and F. N. Ferreira

159

Scientific Visualization as a Tool for Power Engineering EducationP. Slavik and F. Hrdlicka

165

5 - Training for Real World Applications

Presentation

Training in Computer Graphics for Entertainment Production: WhatFuture TDs Need to KnowJ. F. Morie

171

x GVE’99 – Coimbra - Portugal

6 - Human-Computer Interaction

Presentation

Human-Computer Interaction and Cognitive Psychology in VisualizationEducationT. T. Hewett

175

Position Paper

The Human-Computer Interfaces Course at ISTM. R. Gomes, J. M. Brisson Lopes and M. J. Fonseca

179

7 - Media Education

Presentation

Media Education at the Bauhaus-University WeimarC. A. Wüthrich

185

Position Paper

Defining Interactive Multimedia Education: Expanding the BoundariesB. Mitchell

191

8 - Courseware

Presentation

Modularity: The Key to Courseware Re-useJ. C. Teixeira

197

Position Papers

Courseware Reuse or Courseware to Reuse?J. C. Teixeira, C. Páris, J. Madeira and J. Brisson Lopes

199

Framework for Development of Course Material for Teaching ComputerGraphicsV. V. Kamat

203

List of Participants 207

Educational Trojan H orses

Mike BaileyUniversity of California at San Diego

San Diego Supercomputer Centermjb@sdsc.edu

I have always found that it is difficult to teach _at_ people. It is very much like throwing abucket of water on your plants: you can't be sure how much has soaked in and how muchhas evaporated or run off somewhere else. You can't even be sure if water is what theyreally need right now.

Rather, I believe it is much more productive to "trick" people into teaching themselves. Aseducators, our job is less to throw information at our students than to motivate and excitethem into wanting to reach out and pull in the information.

The question then is how to do this. How do we instil educational motivation in someone sothat it blossoms into intrigue and curiosity? How do we make it strong enough for them towant to reach out and figure out more? I call this an "Educational Trojan Horse" approachbecause you slip some education by the students when their guard is down. After all, thebest education happens when people think they are exploring, not being educated.

I believe that computer graphics is _the_ key element of an Educational Trojan Horse.Graphics is such a perfect mix of technology and fantasy that it can be used to excitestudents as no other medium can. The video game appearance causes students to let downtheir guard and allow themselves to be drawn in. The ability to control, design, and simulateleads them into learning. And, they not only develop their own explanations, they aremotivated to discuss it with each other.

This talk will highlight some key technologies and technology trends that can be used tocreate Educational Trojan Horses. It will include 3D graphics, hardware, software, web-based graphics, digital libraries, and common formats.

Strengthening Visual Skills by Recognising Rendering Algorithms

Rosalee WolfeDePaul UniversityChicago, Illinoiswolfe@cs.depaul.edu

Abstract

Computer science students contemplating a career ingraphics need to develop a visual sense, but traditionalcourse topics do not meet this need. Visual analysis isa teaching technique developed for computer scienceinstructors that helps impart this ability. Through theuse of a few visual cues, students learn to visuallyidentify surface algorithms, shaders and lightingtechniques. An interactive software package calledTERA (Tool for Exploring Rendering Algorithms)provides nearly a million image combinations thatstudents can use to practice their visual identificationskills.Keywords: algorithm visualisation, computer graphicseducation.

1. Background

For a successful career in computer graphics, computerscience majors need to develop a visual sense inaddition to their technical knowledge. Computerscience graduates hired by the graphics industry willwork side-by-side with artists, and it is essential foreffective communication that they have an appreciationand enthusiasm for the visual aspects of the field. Inaddition to programming skills and knowledge ofoperating systems, recent labor market analyses list“strong visual style sense” and “understanding howartistic elements work with technical elements,” asdesirable traits for programmers and technical directors[1,2]. An increasing number of graphics houses wantto see a demo reel or a portfolio from computer sciencegraduates who count Unix shell scripting, C++programming and the mathematical derivation ofNURBS in their skill set.

There are many exemplary resources [3, 4] andlearning tools that allow students to examine aspects ofgeometry [5, 6, 7, 8, 9, 10], rasterisation [6, 11],illumination [6, 12, 13, 14] and the entire 3Drendering pipeline [13]. These present visualisationsthat a student can use to explore and study an algorithmin isolation. This is a sound pedagogical approach fora first introduction, but eventually a student will needto understand an algorithm’s behavior in co-operation

with other algorithms and in the context of a finishedimage.

Unlike other specialities in computer science, thechoice of a graphics algorithm usually cannot be basedsolely upon space, speed and implementationconsiderations. Equal in importance to time andmemory requirements is the visual effect that agraphics algorithm produces. In fact, visualappearance will be the overriding factor for someapplications. Visual understanding is essential todeveloping and debugging new algorithms. Forexample, when writing a shader, a programmer looks ata rendered sample, analyses how its appearance differsfrom the ideal and uses the information thus gained torefine the shader. An essential aspect to developing avisual sensibility is the ability to identify and comparerendering algorithms. Visual knowledge can preparestudents to answer such real-life questions as,“Algorithm X gives us exactly the effect we want, butit is prohibitively expensive. Can we get a similareffect by tweaking our implementation of AlgorithmY?”

Although the graphics industry desires a developedvisual sensibility in new computer science graduates,there is no place in the traditional curriculum wherestudents can gain this knowledge. The discipline ofcomputer science draws on mathematics andemphasises algorithm study and development, whichare almost entirely text-based and involve littlevisualisation. Computer science instructors often viewthe prospect of discussing the visual aspects ofcomputer graphics as a daunting, if not overwhelmingendeavor. Typical reactions include

I’m not an artist! How am I supposed to gainthe requisite background to teach this?

I don’t have time for this in class – I have toomany algorithms to cover as it is.

Both reactions are understandable and justifiable.Computer science instructors are not artists and theirgoal does not include producing artists. Further, thereis an enormous amount of technical knowledge in thecomputer graphics discipline, and attempting to decide

what topics belong in an introductory course has beenthe topic of many papers. Any approach to theproblem of inculcating a visual sensibility has to beone that does not take up much lecture time andconstitutes a teaching method that computer scienceinstructors are willing to try. This paper proposes thatvisual analysis is one approach that fits theseconstraints.2. Visual Analysis

Visual analysis is a teaching technique developedspecifically for computer science instructors thatimparts this knowledge [15]. The technique stems fromcritical analysis, which establishes a structure for

examining works of art [16]. Students in the visual artslearn to describe and compare works in terms ofdesign, concept and media.

Instead of design, concept and media, visual analysisteaches students to recognise a small number of visualcues, including:

• visibility of polygon faces or polygon edges• transition from light to dark on diffuse-reflecting

surfaces• color and shape of specular highlights• presence of transparency, reflection, refraction,

patterns and textures

1. First, determine the surface algorithm.Appearance Algorithm

Outlined polygons. Object’s far sides are visible. Uninterrupted horizonlines.

Wireframe

Outlined polygons. An object’s far sides are not visible. Occluded objectsare invisible.

Hidden-line removal

Filled-in polygons. Presence of refraction in transparent objects. (Presenceof reflection, shadows are also helpful)

Ray tracing

Filled-in polygons. No refraction, reflection or shadows. Z-buffer

2. Determine the shader.For Z-buffer

One color per object. Objects appear flat, as if cut from paper. ConstantOne color per polygon. Objects now appear to have a shaded contour. FacetedSmooth transitions from light to dark. Specular highlights follow polygon

edges.Gouraud

Smooth transitions from light to dark. Specular highlights are white,compact and elliptical. “Shiny plastic” look.

Phong

For ray tracingOpaque, colored surface. No highlight. DiffuseOpaque, colored surface with highlight. Diffuse and SpecularTransparent object that appears to bend light. TransparencyPart of scene is visible in the surface of the object. Reflection

3. Determine additional surface interest (both z-buffer and ray tracing)Image appears to have been pasted or glued onto object. 2-D texture mappingObject appears to have been carved from a solid substance like stone or

wood.3-D texture mapping

Object surface appears rough or wrinkled, but its profile is smooth. Bump mapping

4. Light sources Harsh shadow / lighter shadow. Low/high ambient Sharply defined shadow. Point light Presence of a penumbra, but degree of shade is constant within penumbra. Area light Color bleeding; deepening of shadow in corners, under furniture. Radiosity

Table 1: Visual cues and algorithms(visual equivalencies omitted).

• sharpness of shadow• interactions between adjacent diffuse reflectors

(color bleeding)

These cues are usually sufficient to identify a renderingalgorithm as Table 1 demonstrates. Informal studieshave shown that both instructors and students find thislist of cues non-threatening and easy to spot in animage. By learning to observe and describe thesecues, students are able to identify rendering algorithms.

3. Classroom Presentation

In the first lecture, the teacher discusses imagesportraying three or four algorithms that produce starklydifferent effects. For example, the four surfacealgorithms of wireframe, hidden-line, ray tracing andz-buffer create distinctive effects and can bedistinguished by the use of only three visual cues. SeeFigure 1. Each week, the teacher adds more algorithms,and by the midterm, the class has examined allcommonly used rendering algorithms. Until themidterm, the teacher emphasises the characteristicvisual behavior of the algorithms. After the midterm,the teacher shows how multiple algorithms can achieveequivalent effects. By the end of the course, studentslearn to identify the surface algorithm, shaders andtypes of light sources.

While discussing a visual cue, the teacher presents twoor three images that demonstrate it. After explainingthe visual cue, the teacher gives the name of thealgorithm that creates the cue. The teacher then showsa short series of images and invites students to identifythe visual cues and then suggest a rendering algorithm.Students should specify the cues first so that they learnto spot them in the context of an image. Anypremature guesses as to the identity of the renderingalgorithm are met with the response, “Cues first!”.

After the first class meeting, discussions on visualanalysis begin with a short series of images that reviewthe rendering algorithms from the previous meeting.Students first name the visual cues and then suggest arendering algorithm.

The methodology requires only five minutes of anhour-long lecture. Such a small amount of time willnot significantly impact the presentation of othertopics, especially if the teacher covers visual analysisduring the last five minutes, when student focus isbeginning to wander. In fact, there is a way ofpresenting this material without using any lecture timeat all, as will be discussed in the next section.

4. Study Materials

Students need to practice visual analysis outside of theclassroom. While posting images on the Web orpointing out examples in textbooks is a start, studentssay that they benefit more from the question-and-answer sessions held at the end of a lecture. Simplylooking at images does not promote active learning.

Having students learn a rendering package helps a bitbecause students can choose parameters and view theresulting images. However it takes a significantamount of time to learn a package, and since thestudents have a priori knowledge of the surface andshading algorithms, this approach does not provide astudent with a means of self assessment. A betterapproach is to provide students with an easy-to-useinteractive tool that can demonstrate any rendering orshading algorithm while encouraging active learning.

TERA (Tool for Exploring Rendering Algorithms) isan interactive program that facilitates comparativestudy of the visual effects of rendering algorithms [17].See Figure 2 for an annotated screen dump of TERA.

polygons

opaquerefraction,reflection,

shadow

outlinedfilled-in

ray tracing z-buffer hidden-line wireframe

Y Y NN

Figure 1: Three visual cues identify four surface algorithms.

A student can choose a scene and select a renderingalgorithm for any object in the scene. Students canpractice visual analysis using TERA. In “Self Quiz”mode, students select a scene, and TERA presents itwith each object rendered by a random algorithm.Students then guess the rendering algorithm for eachobject in the scene. TERA responds with “Correct”,“Try Again” or “Close Enough”.

The “Close Enough” response is for those cases wheremultiple algorithms produce similar visual effects. Forinstance, Gouraud and Phong shading produce similareffects when no highlight is present.

Always available is the “Tell me more” button, whichprovides specific feedback about a student’s lastalgorithm selection. When a student is in “Explore”mode, the “Tell Me More” button will activate a popupwindow describing the relevant visual cues. In “SelfQuiz” mode, when students get a response of “CloseEnough”, they can click on “Tell Me More” for anexplanation. A pop-up window will list the algorithmthey picked, the actual algorithm and a specificexplanation of why the two algorithms producedeffects that are visually equivalent. For example, if aflat surface is very evenly lit, then Gouraud shadingmay produce variations in shade that are so subtle thatthe result looks like constant shading. In this situation,

if a student picks “constant” and receives the “CloseEnough” response, the “Tell Me More” button willactivate a pop-up window containing the relevantexplanation. See Figure 3.

The new version of TERA is capable of creating nearlya million images for students to analyse. The imagescover surface algorithms, z-buffer shaders, ray tracingshaders, texture mapping, bump mapping, lighting andvisual equivalencies.

Some teachers do not discuss visual analysis in theclassroom, but give a short demonstration of TERAand tell students that TERA is available in the lab.TERA has enough appeal that students are drawn to it,and they spend enough time with it that they can scorereasonably well on a visual identification test [18].

5. Results and Feedback

In my experience, visual analysis adds excitement anda sense of the “big picture” to introductory graphicscourses. Students may not be able to implement everyrendering algorithm when they leave the course, butthey will be able to recognise them and know theirnames, which provides a starting point for furtherinvestigation.

Figure 2: A sample TERA session.

Selectedobject

Algorithmmenu

ImageSelectionmenu

A visual sensibility, fostered by visual analysis,enhances a depth of knowledge of those renderingalgorithms that students do implement. Because theyare already familiar with the visual behavior of thealgorithm, they have expectations of how an imageshould appear and can perform more of theirdebugging on their own without outside help. Studentsno longer ask, “Is it right?” They state, “There’ssomething wrong with my highlight.”

A very exciting development is the favorable feedbackfrom digital artists. This approach promises to help artstudents “grasp the differences between light models,surface algorithms and shaders” [19]. Further workincludes the exploration of developing the approach toexpand the possibilities of better communicationbetween digital artists and computer scientists.

References

[1] Regan and Associates. A Labor MarketAnalysis for the interactive Digital MediaIndustry: Opportunities in Multimedia.Sunnyvale, CA: North Valley PrivateIndustry Council (NOVA), 1997.

[2] Public Affairs Coalition of the Alliance ofMotion Picture and Television Producers,Making Digits Dance: Visual Effects andAnimation Careers in the EntertainmentIndustry. Sunnyvale, CA: NOVA PrivateIndustry Council, 1997.

[3] Owen, G. Using Hypermedia to TeachComputer Graphics. Presented at theGraphics and Visualization Workshop inEurographics ’93 Barcelona, September,1993.

[4] G. S. Owen. Integrating World Wide Web

Figure 3: TERA in “Self Quiz” mode.

Technology into Courses in ComputerGraphics and Scientific Visualization.Computer Graphics 29(3) pages 12-14,August 1995.

[5] D. Schweitzer. Designing InteractiveVisualization Tools for the GraphicsClassroom. In Proceedings of the Twenty-third SIGCSE Technical Symposium onComputer Science Education. pages 299-303, 1992.

[6] P. Min.www.cs.princeton.edu/~min/cs426/applets.html

[7] A. Shabo, M. Guzdial and J. Stasko.Addressing Student Problems in LearningComputer Graphics. Computer Graphics29 (3) pages 38-40, August 1996.

[8] R. Klein, F. Hanisch, W. Strasser. Web-Based Teaching of Computer Graphics:Concepts and Realization of an InteractiveOnline Course. Computer Graphics Annualconference Series, Conference Abstracts andApplications. pages 88-93, 1998.

[9] Y. Zhao, J. Lowther and C. Shene. A Toolfor Teaching Curve Design. In Proceedingsof the Twenty-ninth SIGCSE TechnicalSymposium on Computer Science Education.pages 97-101, 1998.

[10] Y. Zhou, Y. Zhao, J. Lowther and C. Shene.Teaching Surface Design Made Easy. InProceedings of the Thirtieth SIGCSETechnical Symposium on Computer ScienceEducation. pages 222-226, 1999.

[11] A. C. Naiman. Interactive TeachingModules for Computer Graphics. ComputerGraphics 30 (3) pages 33-35, August 1996.

[12] Ansari, M. Simple IlluminationVisualization. Department of Computer

Science, DePaul University, Chicago, 1993.[13] E. Wernert. A Unified Environment for

Presenting, Developing and AnalyzingGraphics Algorithms. Computer Graphics31 (3) pages 26-28, August 1997.

[14] D. Gould, R. Simpson and A. van Dam.Granularity in the Design of InteractiveIllustrations. In Proceedings of the ThirtiethSIGCSE Technical Symposium on ComputerScience Education. pages 306-310, 1999.

[15] A. Sears and R. Wolfe, Visual Analysis:Adding breadth to a computer graphicscourse. In Proceedings of the Twenty-sixthSIGCSE Technical Symposium of ComputerScience Education. pages 195-198, March1995.

[16] R. Arnheim. Art and Visual Perception: APsychology of the Creative Eye. 2nd ed.Berkeley: University of California Press,1974.

[17] R. Wolfe, A. Sears. TERA: an interactiveTool for Exploring Rendering Algorithms.Journal of Computing in Small Colleges 10(4) pages 41-46, February 1995.

[18] R. Wolfe, S. Grissom, T. Naps and A. Sears.A Tested Tool for Teaching ComputerGraphics. Journal of Computing in SmallColleges 12 (2) pages 70-77, November1996.

[19] D. Eber. Personal communication,December 1997.

An Introductory Cou rse on Visualization

Beatriz Sousa SantosDepartamento de Electrónica e Telecomunicações, Universidade de Aveiro

Aveiro, Portugal.bss@inesca.pt

Abstract

A Visualization course offered twice (1997/98 and98/99) as an elective in the MSc degree onElectronics and Telecommunications at theUniversity of Aveiro is presented. Its contents,bibliography and teaching methods are described.Some difficulties encountered during the preparationand lecturing of this course are identified.

1. Introduction

Taking into consideration that Visualization isbecoming very important and useful in many areas,an introductory course on Vizualization seems avalid contribution to the curriculum of any post-graduation in science or technology and thus it wasconsidered adequate as an elective course of the MScin Electronics and Telecommunications offered at theUniversity of Aveiro. This post-graduation programincludes several courses and a thesis and aims to be alarge spectrum degree encompassing mainly one offour areas of Electrical Engineering (Electronics,Telecommunications, Signal Analysis and Processingand Computer Science); this means that it can giveformation either to future consumers or facilitators ofvisualization techniques and systems. This is thecontext where the referred introductory course wasmeant to exist; thus its general objective is tointroduce the students to a new area which evolvesrapidly, addressing the fundamental concepts,providing a basic foundation, good enough to allowand encourage them to apply or proceed work in thatarea. Stating this objective more specifically, thiscourse should introduce the students to whatVisualization is and can do, as well as it should makethem appreciate its benefits and how current toolscan be exploited in many application areas; it intendsto give a consumer’s perspective as well as afacilitator’s perspective. However such a courseshould also serve other more general objectivesrelated to training students to be able to perform thekind of work a post-graduated is expected to do; thisinvolves training them on doing bibliographicresearch, using adequate working methods andcorrect technical communication (i.e. writing andspeaking).

In the following sections a brief description of thegeneral contents, bibliography and teaching methodsused twice (1997/98 and 98/99) on the introductorycourse on Visualization will be presented. Somedifficulties encountered by the author during thepreparation and lecturing of this course are alsoidentified.

2. General Contents

This course was designed to have 36 hours (24lectures) spread along a semester; some of theselectures are dedicated to other activities as justified insection 3. The list of addressed topics, presentedbellow, indicates the duration, focus and specificbibliographic references (briefly described in section4) which have been used.1- Introduction: definitions, history, goals and

principles of Visualization (1 lecture)2- Overview of Visualization applications (2

lectures)3- Human Visual System (1 lecture)4- Visual Representation of Quantitative

Information (1 lecture)5- Framework (1 lecture)6- Visualization Techniques (4 lectures)7- Foundations on the enabling technologies:

Computer Graphics, Digital Image Processingand Human Computer Interaction (7 lectures)

8- Data Characteristics (1 lecture)9- Visualization Products (1 lecture)

In the first lecture, corresponding to topic n.1, ageneral introduction to the area is made. Besidespresenting the definitions and goals of Visualization,a focused approach to visualizing data is presented asa means to select adequate techniques to allowexploration, analysis and then presentation of data.The main references used are the books by Earnshawand Wiseman, and Keller and Keller.This first lecture is of great importance, it is intendedto motivate the students and establish the focus and“philosophy” of the course. Two fundamental ideasare conveyed: a) Visualization is concerned withexploring data and information graphically as ameans of gaining understanding and insight into thedata (not just presentation graphics!) b) how to do

visualization? There is a need of a methodology forselecting visualization techniques, however no“recipes” exist. The focused approach described byKeller and Keller in the first section of their text ispresented. To conclude, a set of images producedfrom seismic data (also from the book by Keller andKeller) is used to illustrate the incremental processone may expect while seeking the best way ofvisualizing data.Topic n.2, is meant to give an idea of the largequantity of applications that Visualization finds in agreat diversity of scientific and engineeringdisciplines. This should make the students aware ofthe importance and usefulness of Visualization, aswell as allow them to better understand all the othertopics of the course. This overview is given throughthe presentation of a large collection of visualizationimages obtained in a variety of disciplines andemphasising the kinds of information revealed ratherthan the details of the visualization techniques. Anumber of examples extracted from severalbibliographic references is used (as the books byBrown et al., Keller and Keller, Brodlie et al.,Earnshaw and Wiseman).The third topic is concerned with the “human part” ofthe visualization process, which is essential. Thecapabilities and limitations of the human visual andinformation processing systems are briefly described,stressing that they have important implications fordesign and that users, in spite of sharing commoncapabilities and limitations, are individuals withdifferences which should not be ignored. The humanvisual system is the main subject and the lecture isbased on a document prepared by the author which isincluded in HyperVis (an on-line tutorial onVisualization) and indicates specific bibliography.In Topic n.4, the fact that visualization is not new inconcept is stressed; most principles that have beenused to produce good maps, scientific drawings anddata plots apply to computer visualization. Thislecture is based on the interesting work of E. Tufte(mainly the 1983 text but also the 1990 text, tosupport colour usage).Topic n.5 presents a framework model describingVisualization systems in abstract terms; thisframework will be used to present techniques, datacharacteristics, products and applications in theremaining lectures. As an introduction to this topic,models of scientific investigation and visualizationprocess are presented. This lecture is based on thecorresponding chapter of the reference by Blodlie etal.The concept of visualization technique, introduced inthe previous topic as the responsible for generatingand manipulating a graphic representation from a setof data and allowing investigation through user

interaction, is developed in topic n.6. The three mainelements of a visualization technique are describedaccordingly to the model used by Brodlie at al.; aclassification and a notation for visualizationtechniques introduced by the same authors is used tosupport the study of a range of techniques, generic innature and which can be tailored to differentapplications. These techniques are illustrated throughseveral examples (extracted also from the referencesby Brown et al., Keller and Keller, among others).Topic n.7 addresses the main aspects of ComputerGraphics, Digital Imaging Processing and HumanComputer Interactions needed to be able tounderstand the visualization techniques. Addressedissues are: a 2D S/W package, geometricaltransformations and projections, specifying a view in3D, colour, overview of the problems and methods toachieve realism, visibility, illumination and shading,histogram operations, filters and image transformsand general principles for user interface design. Theexplanation of these subjects has always in mind itsuse in Visualization and examples includingVisualization techniques already introduced are usedas much as possible.Topic n.8 is concerned with data. A classificationthat seems suitable for describing different types ofdata identified within the used reference model ispresented. An overview of data formats and datacompression techniques is also presented; thisoverview can be supported by the reference ongraphics file formats by Brown and Shepard.A Visualization system is presented, ideally, as anintegrated whole providing means to support theeffective exploration of complex data. Topic n.9presents a classification of visualization software anda variety of existing software products. The adoptedclassification is an hybrid between the classificationused by Brodlie et al. and the one proposed byBrown et al.. The software buyer’s guide included inBrown et al. is used to illustrate the great variety ofexisting commercial products and relevantcharacteristics which need to be considered in thechoice of a S/W product. The study of this subject iscomplemented by demonstrations some of thestudents have to make on several of these products(as part of their assignments).This is the general contents of the curriculum for thepresented course as it has been applied on bothacademic years of 1997/98 and 1998/99. From theinitial proposal of this course [1], the main changecorresponded to a substantial increase on the numberof lectures devoted to the enabling disciplines. Theauthor had initially planned to spend a lesser amountof time on these subjects since elective courses onthose subjects (Digital Image Processing, ComputerGraphics and Human Computer Interaction) are

currently offered at the Department and it seemedreasonable to expect that students taking theVisualization course would have already somebackground on those subjects. The fact that none ofthe students, in both academic years, had thatbackground came as a surprise; in fact it seems that,most of the students had made this choice in order toget some formation in those areas. For this reason,the author decided to increase substantially theamount of time devoted to enabling technologies;this option, although controversial from the point ofview of a visualization curriculum, appeared to bethe most reasonable choice for those students,possibly allowing them to benefit the most from thecourse.The current contents of the course follows, inessence, what has been done by the community ofVisualization educators [2,3,4]. It is expected toevolve, according to the experience obtained, eachtime it will be taught; however the author expectsevolution to occur mostly in the specific way eachtopic is addressed, bibliography, sequence orduration of different topics, rather than on the overallstructure.

3. Teaching Methods

As referred, the course is organised in 1h 30minlectures along a semester, i.e., two lectures a week;its distribution by the topics is indicated in theprevious section. There are no practical classes;students are supposed to perform their assignments,under supervision of the teacher, but on their owntime and with a high degree of autonomy. Thesestudents usually work full time (at the University orat enterprises), which makes very difficult toorganise working sessions involving all the students;that is the reason why only 19 lectures are assignedto the topics, i.e., 5 are left open and are “sacrificed”to activities that could, otherwise, be performed inextra hours. These sessions have been devoted to:• presenting some selected SIGGRAPH videos1

with examples of applications (1 lecture)• visiting facilities where people use visualization

regularly in their work (1 lecture)• presenting student’s assignments (2 lectures)• lectures by people that use or make visualization

(1 lecture)

1 from SIGGRAPH Video Review Issue 108 - Scientific

Visualization 95 which, in spite of having some years, stillcontains interesting examples; however a more recent issuealready exists: SIGGRAPH Video Review Issue 124 - SIGGRAPH97: Visualization Program, that could be interesting.

Until now two visits were organised, both related toEarth Sciences and inside the Campus, one to aLaboratory using Remote Sensing data and anotherto a Laboratory of Geophysics including some fielddemonstration on the acquisition of GroundPenetrating Radar data. Both visits where guided byresearchers from those laboratories, showing thestudents which kind of work they are involved in andhow they use at the moment, and they would like tobe able to use, visualization techniques and tools.Some other laboratories exist at the University thatcan be visited in years to come, as in Chemistry andEnvironment. The two lectures by external people that have beenorganised were concerned with visualization ofArchaeological data and Geographical InformationSystems; however a great variety of other subjectsmight be interesting. As referred, general objectives of this course includetraining students on doing bibliographic search, usingadequate working methods and correct technicalcommunication (i.e. writing and speaking). Sincethese objectives are usually easier to achieve whenthe evaluation is based on practical assignments andthe expected number of students in elective courseson this MSc is low, that was the type of evaluationused in both academic years (the course had 5students in 1997/98 and 6 in 1998/99). Each student has two different assignments, the firstconsists in giving an oral presentation of ≈ 40minutes on some subject and the second, a moredemanding assignment, consists in a work which ingeneral involves some implementation (eitherdeveloping code in some language as C++ or using avisualization S/W as IDL). Both assignments areproposed by the teacher taking into consideration theprofile of each student (background, interests, etc.);however the students are told that they can proposetheir own assignment by presenting an extendedabstract; these proposals may, of course, be rejected.On other courses, offered by the author of this work,some interesting proposals have been presented andaccepted; however, on the Visualization course noneof the students decided to do it, until now. The choice of the subjects for the oral presentationsas well as for the second assignment is related to thecurrent interests of both the teacher and the studentsas well as to interests of other people (from theUniversity or outside) that ask for collaboration insome specific visualization tasks. Along these twoacademic years the oral presentations have consistedmainly in S/W demonstrations (Matlab, IDL, IrisExplorer and OpenGL) or have addressed volumerendering algorithms. Second assignments have beenrelated to the following interest areas:

• visualization of archaeological data from a XVcentury shipwreck existing in the Aveiro Lagoon

• visualization of geophysical data obtained fromGround Penetrating Radar

• visualization of data related to the use ofLocation-Routing models of obnoxious facilities

• development of an hardware processor forvolume rendering implemented with FPGAs

• development of evaluation methodologies forvisualizations used in the study of epilepsy andinvolving data obtained from differentmodalities as surface electroencephalogram(EEG), depth electro-encephalogram (SEEG),tomography (CAT) or magnetic resonanceimaging (MRI)

• development of evaluation methods forvisualizations obtained using raycasting

• development of evaluation methods involvingmodels of Human Visual System

For the supervision of many of these assignments theauthor had the invaluable collaboration of colleaguesworking in enabling technologies and applicationareas.These assignments consist, in general, of abibliographic research on the subject, a proposal ofwhat work could be interesting to develop, animplementation of part of this work, a small technicalreport and a document written as a scientific paper.Most students put a lot of work and enthusiasm intothese assignments and supporting them has been veryinteresting, however students typically take long timeto finish their tasks. In spite of the fact that thiscourse is not a research oriented course such as theone described by Banks [5], many of therecommendations and difficulties presented by thisauthor are also applicable; the choice of subjects forthe second assignment, its supervision and finally thepreparation of the paper corresponds to a significantburden associated to the course; this solution is onlyviable for a very small number of students.This type of assignments have resulted verystimulating for the more interested students, howeverfor less motivated students it is perhaps toodemanding and may result frustrating both for thestudent and the teacher. For these reasons, the authoris considering the possibility of maintaining thissecond assignment for volunteer students and givethe alternative of an examination for the ones whoare not so committed to visualization.One of the main difficulties has to do with thewriting of the “scientific paper”, many students justdon’t like to write (and don’t have enoughexperience) and a special attention is needed to thispart of the work (as well as in the preparation of the

oral presentation). Difficulty in writing is a generalproblem of our students, so some extra effort hasbeen devoted to it, in this course. During a lecture,the teacher usually stresses the importance ofcommunicating correctly any work and generallyinstructs the students how to do it before they startpreparing their oral presentations or their papers;some bibliography on the subject of writing andspeaking is also recommended [6,7]. The teacher alsoencourages students to write in English since it is the“lingua franca” in technology. The process ofpreparing the papers is iterative and some of themneed as many as 3 or 4 iterations. Since the teacherand, in some cases, other people also have to investeffort, not only in the supervision but also inreviewing the paper they become co-authors; whenthe final version of a paper is ready it can bepublished as an internal note (in the Department’sJournal), submitted to a conference or not used at all(accordingly to the nature and quality of thedeveloped work and achieved results).From the 5 papers of 1997/98, four were published asinternal notes; modified versions of two wereaccepted at conferences organised in Portugal andone will be submitted to another conference. One ofthe assignments involved porting to a differentplatform an already existing S/W developed bysomeone who was far away and no longer workingon the subject; in spite of the fact that the student hasdone enough work to succeed in the course, theobtained results did not justify their publication. Thisis always a risk of this kind of assignments and careshould be taken to minimise it, however the authorfeels it is impossible to completely eliminate thepossibility of “something going wrong”.Papers accepted in conferences should be presentedby the students whenever it is possible; thisopportunity should be viewed by students as a“reward” for having done a good work. From theinstitution perspective this should be considered as avaluable contribution to the overall quality and goodname of the degree. For this reason it was possible,until now, to find institutional financial support forthis activity, with the reasonable constraint ofsubmitting papers only to conferences which implylow travelling expenses.

4. Bibliography

Knowing how to search for and use bibliography isan important part of any course, however at MSclevel it is fundamental. At this level lectures aremainly meant to give the basic underlying theory andpoint out important issues, not to present them indetail; for these reasons a good bibliography is in factfundamental, thus a select bibliography is given to

the students and the importance of having thecapacity of obtaining new and adequate bibliographyis stressed.Due to the fact that Visualization is a relatively newdiscipline, no text books seem to be available (atleast in the sense as they exist in other longerestablished disciplines as Computer Graphics orDigital Image Processing). To overcome this type ofdifficulty usually two alternatives exist: use severalbooks and papers or write some notes to support thecourse. The second alternative, in the view of theauthor, is not interesting at MSc level, the studentsought to consult several bibliographic references;however an experienced teacher may write somenotes conveying his/her perspective on some of thetopics. Since the author does not have enoughexperience in lecturing this course, even writingsome notes seemed out of the question. Thus thebook by Brodlie et al. was used as general supportfor the course. As it is stated in its preface:“ this book proposes a framework through whichscientific visualization systems may be understoodand their capabilities described. It providesoverviews of the techniques, data facilities andhuman computer interface, that are required in ascientific visualization system.... the ways in whichscientific visualization has been applied to a widerange of applications are reviewed”; and it seemedcomprehensive enough to be used as a “guideline”. However since it is not a recent and detailed text,some other more up to dated or more detailedreferences were used as support for several topics asalready mentioned in section II. Attempting also toprovide the students with a bibliography that can beuseful in their future activities, the author selected asmall set of bibliographic references that covers thesubjects addressed in the lectures. In the nextsections the general usage of these bibliographicreferences is described and a commentedbibliography is provided. A list of some otherreferences which have been useful, for instance tosupport practical assignments, is also provided alongwith brief comments.During the lectures, but mainly during thepreparation of their assignments, students arestrongly advised to become familiar with otherpossible sources of interesting references such asjournals and conference proceedings as well as thedifferent methods of getting them (as in libraries,throughout data bases, Internet). In fact, theUniversity Library provides an information systemwhich allows the remote bibliographic search in itsdatabases, other portuguese databases and in servicesas INSPEC, which students are urged to use for theirbibliographic research.

4.1. General usage of the selected bibliographyThe first bibliographic reference (of the list presentedin 4.2) is, as already mentioned, the one that willprovide general guidance for the course. Thefollowing three references of the same list can alsobe used as support for several topics, as definitionand goals of Visualization, overview of visualizationapplications, techniques, data characteristics,visualization products and case studies.The most interesting reference by Tufte can be usedto support the study of the history and principles ofVisualization (independently from using computers).The reference on graphics file formats can be usedfor the overview on the relevant issues to file formatsand data compression.The next five references can all be considered as textbooks of the so called enabling technologies ofVisualization: Computer Graphics, Human ComputerInteraction and Digital Image Processing; thus theycan be used by the students to obtain a backgroundon those technologies.Finally, the last reference can be used to support theintroduction to the Human Visual System.

4.2. List of main bibliography• Brodlie, K., L. Carpenter, R. Earnshaw, J. Gallop,

R. Hubbold, A. Mumford, C. Osland, P.Quarendon, Scientific Visualization, Techniquesand Applications, Springer Verlag, 1992

This book was written to be a reference guide for theVisualization community on the technical aspects. Aframework is described and used to presenttechniques, data characteristics, products andapplications. In spite of not being up to dated incertain aspects it still gives a good overview of themain issues that should be addressed in anintroductory course on Visualization.• Brown, J., R. Earnshaw, M. Jern, J. Vince,

Visualization, Using Computer Graphics toExplore Data and Present Information, JohnWiley, 1995

Provides background on the field of visualizationgiving an overview of design issues, visualizationmarket and various visualization products. Itillustrates a wide variety of real-world applicationsthrough case studies.• Keller, P., M. Keller, Visual Cues, IEEE

Computer Society Press, 1993It is intended to people confronted with the problemof discovering the meaning of their data sets. Usingpractical examples from many disciplines, itillustrates visualization techniques, tips and rules ofthumb, that help to produce informative images.• Scott Owen, G. et al., HiperVis-Teaching

Scientific Visualization Using Hypermedia, (on-

line), ACM SIGGRAPH Education Committee,http://www.education .siggraph.org, 1996

It is a hypermedia document under developmentwhich addresses the fundamental topics of ScientificVisualisation.• Tufte, E., The Visual Display of Quantitative

Information, Graphics Press, 1983It is concerned with the design of statistical graphicsas well as with how to communicate informationthrough the simultaneous presentation of words,numbers and pictures. It reviews the graphicalpractice in the last two centuries and seeks to accountfor the differences in quality of graphical designs.• Brown, C., Graphics File Formats: reference

guide, Manning Publications, Prentice Hall, 1995It is a comprehensive guide to file formats used incomputer graphics and related areas (includingvisualization). It discusses implementation anddesign of file formats focusing on the basic issues forits evaluation and development (as data types,organization and compression).• Foley, J., A. van Dam, S. Feiner, J, Hughes,

Computer Graphics: Principles and Practice, 2nded., 1990

This is considered the standard reference inComputer Graphics. It deals with the fundamentaltopics of this area in adequate depth, as well as withmany others.• Watt, A., F. Policarpo, The Computer Image,

Addison Wesley, 1998An updated coverage of subjects from the three fieldsof computer imagery which have previously onlyappeared in separate texts from Computer Graphics,Image Processing and Computer Vision in a coherentoverview.• Shneiderman, B., Designing the User Interface-

Strategies for Effective Human ComputerInteraction, 3rd ed. Addison Wesley, 1998

Provides a complete and current introduction to userinterface design. It offers practical techniques andguidelines taking also great care to discussunderlying issues and to support conclusion withempirical results.• Dix, A., J. Finlay, G. Abowd, B. Russell, Human

Computer Interaction, 2nd ed., Prentice Hall, 1998It is a text book in its area, providing amultidisciplinary perspective of the subject. It coversthe basic psychology and computer technologyinvolved and the interface between them, as well asusability and more advanced topics.• Gonzalez, R. C., R. E. Woods, Digital Image

Processing, Addison Wesley, 1992It is commonly used as a text book in its area; itcovers the fundamental concepts and methodologiesfor Digital Image Processing.

• Fishler, M., O. Firschein, Intelligence, The Eye,the Brain and the Computer, Addison Wesley,1987

Uses an integrated approach on human and machineintelligence, using knowledge from several areas ascomputer science, cognitive science, linguistics,biology anthropology and psychology.

4.3. Other bibliographyOther bibliographic references have been used tosupport either specific topics of the lectures orpractical assignments. These references includepapers published in journals and proceedings ofconferences as well as books. A list of some of thebooks (the ones the author considers may be mostuseful to a reader of this work) is included.• Tufte, E., Envisioning Information, Graphics

Press, 1990Presents a collection of exemplary designsrepresenting all types of information, widelydistributed in time and space, through which designexcellence in complex data representation isidentified and explained. Includes an interestingchapter on how to use colour.• Earnshaw, R., N. Wiseman, An Introductory

Guide to Scientific Visualization, SpringerVerlag, 1992

A book intended for readers new in the field, gives aquick summary of what Scientific Visualization isand can do. Can be indicated as an easy to readintroduction to the subject.• Travis, D., Effective Color Displays: Theory and

Practice, Academic Press, 1991Addresses perception, displays and colour models.Provides practical guidelines to the effective use ofcolour, without disregarding theoretical foundation.It is a useful text for designers wanting to use colour.• Kaufman, A., Recent Trends in Volume

Visualization, 2nd IEEE EMBS InternationalSummer School, June 1996, Berder Island,France, 1996

Provides a survey of volume visualization and itstrends with a focus on biomedical applications.• Rosenblum, L., R. Earnshaw, J. Encarnação, H.

Hagen, A. Kaufman, Sklinenko, G. Nielson, F.Post, D. Thalman, Scientific VisualizationAdvances and Challenges, IEEE ComputerSociety, Academic Press, 1994

Demonstrates techniques, examines diverseapplication areas and addresses relevant issues inScientific Visualization.• Grinstein, G., H. Levkowitz (eds.), Perceptual

Issues in Visualization, Springer Verlag, 1995Addresses issues in the field of applied perception,provides a portrayal of the problems that can be

addressed towards increasing the effectiveness ofinformation displays.• Lichenbelt, B., R. Crane, S. Naqvi, Introduction

to Volume Rendering, Prentice Hall, 1998Provides an introduction to volume renderingconcepts and presents an organised logicalprogression through the volume rendering pipeline; italso contains a CD-ROM with source code whichresults very useful for practical assignments in thesubject.• Murray, J., W. VanRyper, Encyclopedia of

Graphics File Formats, O’Reilly Associates, Inc.,1994

Provides detailed technical information on nearly 100file formats; it also includes chapters on graphics andfile format basics.

5. Conclusions

Offering this course has not been an easy task for theauthor, which have encountered several difficulties;she hopes this work may give a positive contributionto whom it may be interested in preparing a courseon visualization.Perhaps the first and main difficulty encountered inthe preparation of the course is related to the lack oftext books offering comprehensive approaches to thefield; this implies not only a great effort in searchingfor adequate bibliography but also (and perhaps moreimportant) a great effort of organisation andsystemisation of the topics to address. This problemcan be tackled with the (direct or indirect) help ofother more experienced educators and working-groups in the area which publish papers, organiseworkshops and maintain on-line sites containinguseful information (as SIGGRAPH, EUROGRAPHICS,AGOCG). After the initial effort of putting together acurriculum that makes sense, it is necessary to decideon the best approach for the kind of students one isexpecting to encounter, to collect and choose thespecific materials to the lectures and last (but notleast) to choose the subjects and type of practicalassignments and the evaluation to be used. All thesechoices have to be re-evaluated every time the courseis offered based on the accumulated experience of theteacher and feedback from the students. The searchfor new up-to-dated bibliography as well as thechoice and preparation of practical assignmentswhich could result interesting have been the greatestconcerns of the author of this work. These concernsare also expected to be the main concerns in the yearsto come since the proposed curriculum is based onthe curricula of similar courses offered in severalEuropean and American Universities and it isexpected to evolve mostly in the specific way eachtopic is addressed, bibliography, sequence or

duration of different topics, rather than on its overallstructure.

References

[1] Beatriz Sousa Santos, “An Introductorycourse on Visualization”, Revista doDepartamento de Electrónica e Telecomu-nicações (DETUA), Universidade de Aveiro,Vol. 2, N.1, Setembro 1997, pp.13, 16

[2] Domik, G.,”Education for Visualization -Activities of the acm-siggraph EducationCommittee”, Proceedings of the ThirdEurographics Workshop on Graphics andVisualization, Maastricht, August, 1995

[3] Scott Owen, G. S., M. Bailey, K. Brodlie, G.Domik, “Issues in Education forVisualization”, Proceedings of IEEEVisualization 95, Atlanta, October,November, 1995, pp. 427-430

[4] Scott Owen, G. et al., HiperVis, ACM

SIGGRAPH Education Committee (on line),http://www.education. siggraph.org, 1996

[5] Banks, D. , “Including Graphics andVisualization Research in a Master’s-levelCourse”, Computer & Graphics, Vol. 21, n.3, 1997, pp.379-387

[6] Beer, D (ed.)., Writing and Speaking in theTechnology Professions, a practical guide,IEEE Press, 1991

[7] Referencing- (In-text-referencing) (on line),http://www.canningcollege.wa.ed.au/reference (Jan, 1999)

Acknowledgement

The author wishes to thank Prof. Gitta Domik, chairof the ACM-SIGGRAPH Education for VisualizationCommittee, for her great help in the initial definitionof the curriculum and bibliography of the presentedcourse, as proposed in 1996.

Restructuring Comp uter Graphics and Visualisation Curricula at IST

J. M. Brisson Lopes, M. R. Gomes, J. C. Bernardo, J. Jorge and J. M. PereiraDepartment of Informatics Engineering, Instituto Superior Técnico,

Av. Rovisco Pais, 1049-001, Portugalbrisson@ist.utl.pt, {mrg, jcb, Joaquim.Jorge, jap}@inesc.pt

Abstract

This paper addresses the restructuring of theundergraduate and graduate programs in ComputerGraphics, Visualisation and Multimedia that wasrecently carried out at Instituto Superior Técnico. Thepaper details the rationale for the changes in bothcurricula and course content and describes the newstructure. The result is a coherent and complete seriesof courses spanning the undergraduate and M.Sc.degrees in Informatics Engineering. It furtherdiscusses the structure of the basic course onComputer Graphics within this context and drawsconclusions to contribute to the ongoing discussion onComputer Graphics, Visualisation and Multimediacurricula.Keywords: course curricula, Computer Graphics,multimedia, undergraduate programs, graduateprograms.

1. Introduction

The Informatics and Computer Science undergraduateprogram at Instituto Superior Técnico (IST) started inthe academic year of 1989/90 under the aegis of threedepartments, Electrical Engineering, Mathematics andMechanical Engineering, which shared the co-ordination of the new degree program. Roughly at thesame time the M.Sc. program in Electrical Engineeringand Computers offered a specialisation in Computing.Courses on Computer Graphics were taught within theundergraduate and graduate programs.

This scenario lasted until 1998 when IST decided tocreate a new department, the Department ofInformatics Engineering (DEI). The decision was takengiven the need for better co-ordination of theundergraduate program which had come of age and thenumber of teaching staff and undergraduate studentsinvolved. The new department was officiallyestablished September 1st, 1998.

DEI´s creation started a new and extremely dynamicprocess. The novel Department’s professors engaged inan overall revision of the department's teachingactivities with three purposes in mind. First, previousexperience had shown an effective need for

adjustments in the undergraduate program syllabi.Second, the new department needed to assess in detailand demonstrate its requirements in terms of teachingstaff. Finally, the revision was necessary because of thecontinued changes in Information Technology.

At the same time this dynamic process hadimplications in the graduate curricula. The result wasthe decision to create a new autonomous M.Sc.program addressing the needs of graduate student inthe scientific area of the new department and relatedareas.

By and large, this constituted a unique opportunity toaddress all these matters from a global perspective.Instead of taking the classical approach and addressundergraduate and graduate levels separately, this wasthe time when the two programs could be changedtogether to make the graduate program the logicalsequence of the undergraduate program.

The outcome was the adoption of a model currentlyadopted in other universities, notably in the UnitedStates. This model allows undergraduate studentsintending to continue their studies through the graduateprogram to take advanced courses from the M.Sc.program while in their senior year. This conceptbecame the Integrated M.Sc. Program.

After restructuring the undergraduate program, allscientific areas in the Department of InformaticsEngineering were asked to set-up their particulargraduate (M.Sc.) curricula. Later on, the integratedundergraduate-graduate scenario was finally revisedand approved. In this process the Computer Graphicsand Multimedia professors of DEI took the opportunityto assess and reprogram all courses within theirresponsibility.

This paper reports the results of that work. Section 2lists the objectives set at the beginning while section 3presents the rationale and describes the new programstructure for the area of Computer Graphics andMultimedia. Section 4 addresses in detail one of thebasic components of the program, the ComputerGraphics course. The final section presents someremarks and general conclusions.

2. Objectives

In Portugal, as in most other countries, the supply ofinformation engineering professionals has not keptpace with job market needs. Besides the hiring frenzydriven by the arrival of the Euro and Y2K endeavours,many new Portuguese companies have been created inrecent years, with some of them involved ininternational activities such as ParaRede and EasySoft,to name a few. This shows clearly that there is a needfor new and highly qualified professionals at all levels.Furthermore, the continuous development ofinformation related technologies, of which the WWWis a good example, requires professional skills thatwere not taught a few years back. Furthermore,currently employed professionals are required tocontinuous update their qualifications in order to be onpar with technological development.

Therefore, the first goal of the new curricular structurewas to produce an undergraduate program that wouldsatisfy the above needs. The second goal was to createa specialisation area in Interactive Systems andMultimedia suited to the needs of future professionals.Another important goal was to set the groundwork fora doctoral program in Interactive Systems andMultimedia.

Since the topics to cover are interrelated, we decided todevelop an integrated approach spanning bothundergraduate and graduate programs, in order to:

• offer students a coherent global program inComputer Graphics, Visualisation, Multimediaand Interactive Systems.

• establish a complete and logical sequence ofsubjects and courses extending from earlyundergraduate to advanced graduate studies.

• allow students intending to obtain a M.Sc.degree to proceed smoothly and quickly tograduate studies by shortening the intermediatesteps required.

• empower students not intending to pursuegraduate studies to achieve a high degree ofproficiency.

• enable students from other undergraduateprograms to enrol in our graduate program.

3. The M.Sc. Program in Interactive Systemsand Multimedia

This section presents the Master’s program inInteractive Systems and Multimedia, its guidelines andstructure, together with the outline of each course’ssyllabus.

3.1. RationaleThis following addresses the general guidelines for thenew program organisation. We will discuss therequirements at undergraduate level before moving onto the graduate program in Interactive Systems andMultimedia.

All undergraduate students must become familiar withthe basic principles of Computer Graphics. Theseminimum requirements comprehend both two-, three-dimensional, surface and hierarchic modelling and useof colour. Furthermore, most undergraduate studentswill take courses on Computer Aided Design andHuman-Computer Interaction. These complete theknowledge requirements at the undergraduate level andare the basis for graduate level studies. Any studentapplying to the master’s program must have the abovebasic Computer Graphics background. If not, they willhave to complete it before moving to the masterprogram by taking the missing courses.

Students at the end of the fourth year of graduationmay elect to proceed directly to the master’s program.Instead of moving on to the fifth and final year of theundergraduate degree, they can take advanced coursesfrom the master’s program first year.

The master’s curriculum in Interactive Systems andMultimedia includes courses from two major areas,Computer Graphics and Information Systems. Studentsintending to specialise on the design, implementationand use of Computer Graphics Systems opt for a majoron Computer Graphics and a minor on InformationSystems. Conversely, students whose interest lies onInteractive Systems can opt for a major in InformationSystems and a minor in Computer Graphics.

Students earn credits from courses in these two areas.Each major begins with a fundamental course offeredduring the first semester of graduate studies. After thatstudents must take at least four courses in their majorarea. Of these four courses, two intermediate curses onthe second semester are required studies for the twoadvanced courses on the third semester.

A master's thesis to be done during the second yearcompletes the graduate master program. The thesismust be supervised by a professor of DEI’s ComputerGraphics & Multimedia scientific area.

Students choosing a major in Computer GraphicsSystems must take the Three-dimensional Modellingand Visualisation course in the first semester. Duringthe second semester, students will take two morespecialised courses, one on Visualisation and the otheron Modelling.

Students choosing a major in Interactive Systems musttake the course on Human-Computer Interaction in thefirst semester. These students will then have to take thetwo more specialised courses on Hypermedia andIntelligent Interfaces in the second semester.

In the third semester, students can choose two coursesfrom the five advanced courses available. These areVisual Languages, Virtual Reality, ScientificVisualisation, Geographic Information Systems andAdvanced Topics on Computer Graphics andMultimedia.

3.2. Program StructureThe courses in the new structure were classified intofour categories: G, L, M and A.

G courses are typical undergraduate courses andconstitute pre-requisites for enrolling into the masterprogram. The Computer Graphics course is part of theInformatics and Computer Science undergraduateprogram. Other related undergraduate programs havethe option for the Computational Graphics Methodscourse, which is a version of the Computer Graphicscourse adapted to such other undergraduate programs.L courses are offered both as part of the undergraduateprogram, and of the first semester of the graduatemaster program

M courses take place in the second semester and aretypical master program courses. Completion of thecorresponding L courses is necessary to take any Mtype course. M courses are pre-requisites for all A typecourses.

Type A courses are final master’s program coursesaddressing advanced topics and take place during thethird semester.

Figure 1 presents a diagram showing the courses andtheir pre-requisites. Table 1 lists the undergraduate andgraduate courses according to type, curricular semesterand year.

3.3. SyllabiThis section presents the abridged syllabi of thecourses listed above. The Computer Graphics courseis presented in more detail in the next section. Thereason for this is the fact that this course is theintroductory fundamental course for the whole area ofComputer Graphics and, therefore, deserves specialattention.

The first semester course on Three-dimensionalModelling and Visualisation aims at teachingstudents how to create and use different 3Drepresentation models, model scenes and animate

objects. The curriculum also addresses visualisationtechniques to cope with specific requirements likegames. This course is project oriented.

Course Designation Type Yr. Sem.

Computer Graphics G 2 2

Computational Graphics Methods G 5 1

3-D Modelling and Visualisation L 5 1

Human-Computer Interfaces L 4 1

Intelligent Interfaces M 2

Hypermedia M 2

Graphical Modelling Complements M 2

Visualisation Complements M 2

Visual Languages A 2

Virtual Reality A 3

Scientific Visualisation A 3

Graphical Information Systems A 3

Adv. Topics in CG and Multimedia A 3

Table 1. Undergraduate and graduate coursesaccording to type, curricular semester and year.

The course on Human Computer Interaction (firstsemester) is intended to teach students how to createand evaluate user interfaces. To achieve these goals,students learn the concepts of Human Factors(cognitive and motor-sensorial), and Usability and howto use and apply different software methodologies andtools for interface creation. This course is projectoriented.

The second semester course on Intelligent Interfacesexplores new paradigms in user interaction that willreplace present-day approaches based on directmanipulation and the desktop metaphor. The coursediscusses multimedia input analysis; recognition ofspeech, gestures, gaze and calligraphic sketches;synergetic combination of multiple interactionmodalities; the architecture and construction of noveluser interfaces; presentation of multimediainformation; agent-based interfaces and naturallanguage.

The course on Graphical Modelling Complementscourse (2nd semester) addresses a large set of topicsrelated to the design and implementation of genericgeometric algorithms in both 2D and 3D. Starting witha comparative study of several solutions to someclassical and elementary geometric problems, it thenprogresses to the description of more complexgeometric modelling techniques. The course thenpresents suggestions for the application of those resultsto solving practical problems in different domains such

as Virtual Reality, Computer Aided Design, ComputerVision, Robotics, etc.

The second semester course in VisualisationComplements is an exposition of state of the arttechniques in 3D rendering such as antialiasing theory,texture mapping, ray-tracing advanced strategies,radiosity methods or volume rendering. The motivationof this course is to explain the many techniques thathave evolved, with sufficient theoretical rigour and inenough detail to enable their use in a project onrealistic image synthesis.

In the third semester, the course on Visual Languagesfocuses on a class of formal languages based on thenon-linear organisation of information. The relevanceof such languages is reflected on the proliferation ofsystems such as spreadsheets, CAD and GIS systems,which structure information in an essentially visualmanner. The course establishes the connection andcontrasts visual to conventional formal languages,balancing a discussion of theory with the study ofpractical applications.

The objective of the third semester Virtual Realitycourse is to provide a comprehensive technical reviewof the engineering details underlying the developmentof virtual environment displays. Students should notonly become familiar with concepts like immersionand stereo rendering but also understand existingvirtual reality systems and their associatedtechnologies. Special focus is given to the Webvisualisation. Students will then explore this topicthrough a project where they must apply VRML andJava technologies.

The advanced course on Scientific Visualisation (thirdsemester) addresses the architectural solutions found insuch systems and the means and ways of achievingrealistic scientific data visualisation applied by bothtraditional and immersive visualisation systems,including real-time visualisation and user interaction.This course is project oriented.

In third semester advanced course on GeographicInformation Systems the students will apply previousknowledge on Organisation Models, InformationSystems, Interactive Systems and Computer Graphicsto study not only the technology (information access,information publication, specific methodologies) butalso to study the impact of technology onorganisations. Case studies will be used.

The third semester advanced course on AdvancedTopics on Computer Graphics & Multimediaprogram will be defined each year and will include thelatest scientific subjects in the field.

4. The Basic Course: Computer Graphics

The Computer Graphics course is the entry point tothe Computer Graphics world to most students enrolledin the Informatics and Computer Scienceundergraduate program. For this reason, this paperaddresses it in more detail in the following.

Note that the Computational Graphics Methodscourse, that is offered at IST to students enrolled in theElectronics and Computer Engineering and AppliedMathematics undergraduate programs, runs parallel tothe Computer Graphics course and addresses roughlythe same subjects. However, it follows a moretraditional pedagogical approach with adaptations tothose programs and in line with those programs' nature.

4.1. Objectives and GuidelinesThe Computer Graphics course is the basic course atundergraduate level in its scientific area. After thiscourse, students are expected to understand and applythe fundamentals and algorithms in ComputerGraphics.

At the same time, this course plays the very importantrole of attracting students to the area of ComputerGraphics. Students are familiar with the latest advancesin the area since they have most certainly experiencedit in computer games and while browsing the WorldWide Web (e.g., VRML). Moreover, what used to beexpensive hardware and heavy applications can befound today in multimedia PCs at a low cost. Mostbasic algorithms come already embedded in low costhardware.

From the above, it is evident that the traditionalpedagogical approach to a basic course in ComputerGraphics is no longer on par with the today's realityand can be a negative factor in keeping studentsinterested in the subject. This is why we decided tofollow a different pedagogical approach that looks atComputer Graphics from a global structuredperspective using today's typical applications and thenproceed to more detailed levels.

4.2. Course ContentsA typical semester consists of thirteen weeks with threehours of lecture time per week at IST. In line with waspreviously said we propose to develop the coursecurriculum of the introductory Computer Graphicscourse as in the following course outline (the numbersin parentheses indicate the number of classes dedicatedto each subject).

1. Introduction and General Concepts [2]2. Geometric Transformations [3]3. 3D Visualisation [7]

3.1. Projections and the Rendering Pipeline3.2. Geometric Modelling and Object Hierarchy3.3. Visualisation Algorithms3.4. Clipping, Scan Conversion, Hidden Line

and Surface Removal4. Parametric Surfaces [2]

4.1. Representation4.2. Tessellation

5. Colour models, halftoning, reproduction [2]6. Shading and Illumination [2]7. Ray-Tracing and Radiosity [2]8. Animation [2]

We will use the classic textbook "Computer Graphicsand Applications" by Foley et. al.[1]. Thiscomprehensive classical textbook has beentraditionally used at IST although it leaves much to bedesired from the pedagogical point of view. We arecurrently looking for alternatives.

4.3. Other Features of the CurriculumThe practical pragmatic aspects of programming andalgorithm implementation are relegated to the weeklylaboratory and practice classes (two hours per week)instead of recitation sessions.

The study and description of 2D windowing systemsand raster primitive operations, already reduced inprevious editions of the course, disappears asautonomous curriculum item. The discussion of two-dimensional primitives appears at its logical place inthe discussion of the visualisation pipeline, i.e., afterthe discussion of the scan-line algorithm conversion.

We introduce students to modelling languages such asVRML to construct a fairly complex scene as their firstassignment. The second laboratory project focuses onthe study, implementation and experimentation ofthree-dimensional visualisation algorithms, building onthe enthusiasm students traditionally vow to suchactivities.

5. Conclusion and Final Notes

This paper presented the new program structure inComputer Graphics and Multimedia of the integratedundergraduate and graduate (M.Sc.) programs inInformatics and Computer Science at Instituto SuperiorTécnico, Technical University of Lisbon. Weaddressed the aims of the novel integrated program anddescribed the curricula of both programs' courses,paying special attention to the introductory course inComputer Graphics.

The new curricula are the outcome of collaborationwork carried out by the professors of the DEI'sComputer Graphics and Multimedia scientific area andfollow the experience gained by similar initiativesalready in place in other countries, especially in theUnited States. In our view, the program and curriculaproposed here constitute a well balanced andcomprehensive set, covering both the fundamentals andadvanced topics in this scientific area in a progressivemanner. Furthermore, the advanced topics proposed arewell in line with current research trends in our field.

The proposed program is also clearly in tune withmodern requirements for competent ComputerGraphics professionals at both undergraduate andgraduate (M.Sc.) levels. Moreover, the programensures a smooth transition from undergraduate tograduate level with minimum delay.

The program will be set in place at the beginning of the1999/2000 academic year and is expected to be in fulloperation by the year 2001.

References

[1] J. D. Foley, A. van Dam, S. K. Feiner, J. F.Hughes, Computer Graphics, Principles andPractice, 2nd. Edition, Addison-Wesley,1990.

Figure 1. Course diagram showing course dependencies.

ComputerGraphics

ComputationalGraphics

Algorithms

VisualLanguages

Virtual Reality

ScientificVisualisation

GeographicalInformation

Systems

Adv. Topics onCG &

Multimedia

Undergraduate 1st Semester 2nd Semester 3rd Semester

Human-ComputerInterfaces

IntelligentInterfaces

Hypermedia

3D Modelling& Visualisation

GraphicsModelling

Complements

VisualisationComplements

©1999, IST & JMBL

Adapting Computer Graphics Curricula to Changes in Graphics

Lewis E. HitchnerDept. of Computer Science

California Polytechnic State Univ.San Luis Obispo, CA U.S.A.

hitchner@csc.calpoly.edu

Henry A. SowizralGraphics Division

Sun Microsystems, Inc.Menlo Park, CA U.S.A.henry.sowizral@eng.sun.com

Abstract

Introductory computer graphics courses arechanging their focus and learning environments.Improvements in hardware and software technologycoupled with changes in preparation, interest, andabilities of incoming students are driving the needfor curriculum change. Past courses focussed onlow- and intermediate-level rendering principles,algorithms, and software development tools. Many ofthese algorithms have migrated into hardware.Though important knowledge for advanced graphicsprogrammers, most graphics applicationsprogrammers have no need to study at this level,much as application programmers have no need tostudy hardware systems or assembly levelprogramming. Courses need to focus onintermediate- and high-level principles, algorithms,and tools. A fundamental need in modern graphicscurricula is integration of a 3D graphics API intothe instruction. This paper presents experiencesteaching this focus with both low and high levelgraphics programming API's. The experiences weregained in courses at an undergraduate universityand in multi-day industrial courses for experiencedprofessional programmers.Keywords: computer graphics, curriculum.

1. The Changes in Directions ofIntroductory Computer GraphicsCourses

The curriculum of courses for the study of computergraphics has changed and continues to change inresponse to technology and student needs. We firstreview graphics technology, students, and curricula,as related to computer graphics instruction during thethree stages of acceptance of a new technology(ignoring the outlier stages): that of the "EarlyAdopters", that of "Early Majority", and that of "LateMajority"[1]. Next a set of principles and goals for anew curriculum for introductory computer courses ispresented. We follow this with examples of practicalexperience in teaching such a course at a polytechnicstate university and in multi-day short courses forpractising professionals.

1.1. Graphics Technology, Students, and CourseCurricula in the "Early Adopters" Era

From the 1970's through early 1980's computergraphics courses existed primarily at largeuniversities. Computer Science education was still inits infancy and courses in computer graphics wererare. They were usually specialty courses at a fewschools that could afford expensive graphicshardware, usually funded by government researchgrants. During that time the curriculum emphasiswas on recent developments in research that had notyet entered the public domain or commercialsoftware worlds. Research focussed on developingnew techniques, more functionality, and algorithmsthat performed more efficiently.

Courses taught during this era assumed the followingabout students and graphics systems:• students were highly motivated technologists

with advanced science and mathematicspreparation (computer science, mathematics,and engineering majors),

• most students had little or no experience withthe new field of 2D and 3D graphics,

• students should learn all the fundamentalprinciples, algorithms, and techniques thatresearchers had recently developed,

• graphics hardware devices were not widelyavailable, and they depended upon specialpurpose software interfaces.

The curriculum of courses during the pioneer daysreflected these assumptions. Typical topics included:

• an overview of graphics technology andapplications,

• introduction to low level graphics hardware:CRT's, input devices, processors, peripheraldevices,

• algorithms for performing basic renderingpipeline operations:• line and circle drawing (e.g., Bresenham's

Algorithm),• line and polygon clipping,

• matrix transformations (primarily for 3Dviewing, though also for modeling),

• lighting and shading algorithms, and• image generation algorithms (polygon scan

conversion, visible surface computation, Z-buffer)

• introduction to 3D modeling: parametric curveand surface generation, some proceduraltechniques,

• software technology for interfacing graphicsprocessors and displays to mini-computers andmainframes using locally developed (one-of-a-kind), or proprietary commercial, softwarefunction libraries.

Most graphics textbooks used during this erareflected these assumptions. University coursesfollowed these assumptions and covered topics withthe same emphasis as stressed in popular textbooks.During this era there was also somewhat of anelitism. Only computer scientist-technologists withadvanced university degrees were capable ofdeveloping (or even using) complex graphics displayprograms.

1.2. Graphics Technology, Students, and CourseCurricula in the "Early Majority" Era

By the early 1980's, graphics technology becamelower priced and more widely available. Graphicsworkstations, such as those made by Evans andSutherland, SGI and other vendors, provided high-powered processors for those schools able to affordthem. Graphics peripheral processors (e.g., the AED512) and early versions of PC graphics systems alsobecame available at somewhat more reasonableprices.

Early attempts at graphics software standardizationoccurred during this era. Some API's were shortlived or never widely used (e.g., the "original"graphics standard, CORE), or were extended andrevised as new needs and features evolved (e.g.,GKS to GKS-3D, PHIGS to PHIGS+). Such movingtargets were not very successful at becoming widelyadopted. Proprietary graphics software packagesdominated the industry.

Students during this era were somewhat similar tothose of the earlier era. However, the popularity ofcomputer science as a career attracted many morestudents from more educationally diversebackgrounds. Many graphics students during this erahad weaker math and science knowledge than earlierstudents. Fundamental topic areas of graphics(geometry, physics of color and lighting, graphics

hardware device operation) were unfamiliar to them,and had to be covered in course curricula. Morestudents had some familiarity with using 2D (andsome 3D) graphics in video games.

Although graphics software was more standardized,educators and textbook authors were wary ofteaching students the "API du jour". They avoidedteaching standards that were likely to be soon out ofdate. Instead, most courseware developers designedtheir own generic, "standard-like" software packagesto use in accompaniment with a particular text orgraphics hardware system. It was also common for auniversity or individual instructor to write his or herown graphics package (often publicly shared withother universities via anonymous ftp in the pre-WWW days). Unfortunately, these "teaching API's"were often minimal sets of functions. They weresufficient for learning the principals of graphicssoftware development, but were not representative ofreal-world, industrial-strength packages. And, ofcourse, knowledge of them was not portable toanother school or to a job. However, since moststudent assignments were small, one-shot programsand never lived long enough to get to themaintenance phase of their life cycle, use of a "toy"API was acceptable. That met the goal of providingthe necessary learning experience. Of course, aftergraduation, the employers of these students had totrain them to use "real world" graphics softwaresystems.

In response to the availability of and interest ingraphics technology, many computer sciencedepartments adopted graphics as a part of theircurriculum. The ACM Curriculum '91 [2] specified agraphics course as a technical elective. The use anddevelopment of graphics applications had nowbecome accessible to any computer, math,engineering, or science student who enrolled in auniversity graphics course.

1.3. Graphics Technology, Students, and CourseCurricula in the "Late Majority" Era

Current, late 1990's graphics hardware is orders ofmagnitude faster and cheaper, and it is much morerobust and fully functioned, than earlier technology.The prevalence of personal computers with low cost,high-performance 3D graphics accelerators, is thedominant technology. However, extreme increases inperformance to price ratios have occurred forproducts at all pricing levels.

Furthermore, graphics hardware technology hasevolved to a state where rendering algorithms are

specialized for high performance. Certain aspects ofgeometry representation, lighting and shading, andinteraction handling are primarily implementedwithin restricted preconditions rather than generalpurpose ones. For example, most modern graphicsrendering hardware either requires all geometry to berepresented using triangles only or else is tuned forhighest performance using triangles. Also, somecurrent 3D graphics hardware systems implement theentire OpenGL rendering pipeline in hardware.

Nearly as great an increase in performance andfunctionality has occurred for graphics software. Themost significant aspect of software evolution hasbeen the emergence and acceptance of standardizedgraphics software API's (Application ProgrammerInterfaces). Today there are several standard 3Dgraphics API's used in industry. The three leading,nearly universally used "low level" API's areOpenGL (evolved from SGI's proprietary GL andnow available on nearly all platforms), Microsoft'sDirect3D (Windows platforms), and Apple'sQuickDraw 3D (Apple platforms). There are alsomany high level 3D API's, though currently there isnot the dominance by one or two as there is at thelow level.

In addition to graphics API's there are four othersignificant factors in graphics software development:

1. Powerful, low cost IDEs (IntegratedDevelopment Environments) for rapid and easycode development and debugging. For example:Microsoft and Imprise (formerly Borland)products.

2. Extensive libraries of general purpose supportfunctions that provide higher level graphicscapabilities and simplified GUI development,most of which are either free or very reasonablypriced. For example: GLUT (GL UtilityLibrary), MFC (Microsoft Foundation Classes).

3. A number of powerful software packages forhigh level, application independent developmentsuch as ray tracing and animation systems(many of which are public domain). Forexample: POV Ray.

4. The World Wide Web and its wealth of free andinstantly accessible demos, software tools, datasets, and human resources (via email,newsgroups, and chat sessions).

Graphics students of today are quite different fromthose in earlier years. Some characteristics are:

1. Most technical university students (i.e.,computer science, engineering, math, or science

students) enter a first course in computergraphics with significant prior exposure tographics concepts, such as lighting, perspective,and 3D viewing and navigation. Many studentshave had significant programming experience,and it is not uncommon for them to have alreadywritten 2D and 3D graphics programs.

2. Although most students come to school withgreater computer skills and graphics experiencesthan their predecessors, on the other hand, manyarrive with greater handicaps. Even amongengineering and computer science students, theirmathematical, problem solving, and logicalreasoning skills appear weaker than in prioryears.

In the opinion of these authors, contemporarycurricula of most university level graphics coursesand textbooks do not appropriately emphasize themost critical aspects of graphics hardwaretechnology, nor have they appropriately adapted tothe backgrounds of their students.

It is no longer necessary to introduce today's studentsto such topics as a pixel, a bit-mapped image, a colorpalette, or basic RGB color systems. For many, evenmore advanced topics such as 3D viewing, 3Dnavigation, and texture mapping are familiar. Avid,even casual, web-surfers or PC gamers of today arefamiliar with using these terms and features.

Many fundamental algorithms and procedures ofgraphics are no longer relevant, even though they arepedagogically valuable. For example, Bresenham'sline and circle algorithm and the polygon scanconversion algorithm are not relevant. Suchoperations are done in very low level hardware (e.g.,microcode or ASIC's) and even there, traditionalalgorithms are not always used. Clipping operationsare also buried deep within the hardware. Cohen-Sutherland line clipping and re-entrant polygonclipping algorithms are good intellectual exercisesfor today's students, but have less relevance tohelping them learn useful graphics softwaredevelopment techniques.

Now that small set of graphics API's are nearlyuniversally used (and also becoming more similar),students should learn to apply the fundamentalprinciples of graphics within these environments, notwithin make-believe, "toy" software environments.A few recent textbooks integrate modern API's, andone of them [3] has been adopted by a number ofuniversities. However, to our knowledge notextbooks fully meet the criteria of presenting

graphics technology in the context of today'shardware and software systems.

Regardless of criticisms of graphics educationweaknesses, the accessibility and usability ofcomputer graphics technology today is trulyphenomenal. No longer are those who study and usegraphics members of an elite club. Nor are theyrestricted to university engineering and sciencestudents. We now have experienced the”democratization” of computer graphics.

2. New Models for Introductory GraphicsCourses: SIGCSE and SIGGRAPHPanels

In response to the need for modernizing the curriculaof graphics instruction, a panel at the recent ACMSIGCSE (Assoc. for Computing Machinery, SIGComputer Science Education) Technical Symposium[4] chaired by one of this paper's co-authors,discussed and presented a proposal for contemporarycomputer graphics curricula. In addition to the panel,a Birds-of-the-Feather session co-sponsored by theSIGGRAPH Education Comm. at the sameconference held discussions following the panel. Asimilar panel with three of the four same panelistshas been submitted and accepted for presentation atthe ACM SIGGRAPH '99 conference (Los Angeles,August 1999).This panel proposed the following philosophy of thefirst graphics course:

• Computer graphics is inherently 3D and coursesshould be also.

• The fundamental subject of a computer graphicscourse is geometry and how it is expressed incomputational terms. Thus, geometry is a majorpart of the introductory course. Geometry isexpressed in terms appropriate to the field, suchas coordinate systems, transformations, andsurface normals. The basic shape is the triangle.

• Computer graphics is intrinsically visual, andeven the most technically oriented graphicspractitioner must be aware of the visual effectsof algorithms. Unlike other areas of computerscience, algorithms must be considered not onlyfor time and memory usage, but also for theirvisual effect.

• Besides geometry, computer graphics is aboutlight and surfaces, and about developingalgorithms to simulate their interplay. Coursesneed to include material about light and surfaceproperties and about the distinction between the

ways various algorithms present light andsurfaces visually.

• Computer graphics has matured to a state inwhich there are a small number of high-levelAPI's that support all the fundamental conceptsneeded for early work. Courses should be builtupon this kind of high-level approach.

• Computer graphics should be interactive.Courses should include interactive projects andcover event-driven programming.

3. Experiences in Teaching IntroductoryGraphics Courses

One co-author has taught undergraduate universitygraphics courses from 1984-1988 at the Univ. ofCalif., Santa Cruz, and from 1995 to present atCalifornia Polytechnic State University (Cal Poly).The syllabus and topics emphasis of courses hasshifted significantly.

The original syllabi were similar to that of the "EarlyMajority" era model, based upon the sameassumptions about available graphics technology andstudents as stated above. The topic focus was onfundamentals (hardware devices, geometry, viewing,lighting, and shading) and algorithms forimplementing the rendering pipeline. Interactionmethods were limited to fairly simple eventhandling, and they used console text i/o without aGUI. Some hierarchical modeling was taught usingmodeling transforms implemented by theprogrammer. Early versions of the course used a"home grown" API (implemented on top of a devicedependent 2D graphics API). Later versionsmigrated to proprietary versions of GL (SGI andHP). Typical programming lab assignments includedimplementation of Bresenham's line drawing,Cohen-Sutherland line clipping, re-entrant polygonclipping, and simple hierarchical modelingapplications.The 1996-98 Cal Poly graphics courses use modernAPI's. The co-author teaches a version that usesOpenGL for the first half of the course, and usesOpen Inventor for the second half [5]. In this coursethe topic focus is also on fundamentals. But,instruction in implementation techniques has shiftedto higher level aspects of rendering, event-driveninteraction handling with GUI's, complexhierarchical modeling including extensive study ofscene graphs, and software design using the twoAPI's. The first lab assignment draws complex 3Dprimitive shapes with interactive control of color,orientation, and line style. A second lab assignmentdraws chairs, a table, and a floor and requiresinteractive control of position and orientation for

each object and for the viewpoint. A third lab usesOpen Inventor to model a 16-jointed robot withinteractive control for joint manipulation. This labalso requires model development using a graphicalscene graph editor without writing program code.

4. The Future: Java 3D as a LearningEnvironment for Introductory GraphicsCourses

1.4. Overview of Java 3D Design Philosophy andFeatures

Java 3D is an API used for writing 3D graphicsapplications and applets. It is a library of basic andutility classes written in the Java language. Java 3Dis platform independent and extends Java's "writeonce, run anywhere" benefits for applicationdevelopers. It also integrates well with the Internetbecause applets written using Java 3D have access tothe entire set of Java classes. A complete descriptionof the Java 3D design philosophy and its features isavailable in published texts and online webdocuments [6,7,8].The Java 3D programming paradigm includes theseprinciples:• Fully object-oriented implementation• Classes are extensible and compatible with all

other Java 2 Platform libraries• Scene Graph based for both geometry and

behaviors• High performance a primary design and

implementation philosophy:• Layered Implementation: Native code

based, aiming at hardware acceleration,• Application programmer can specify what

will change so that system can performoptimization,

• Supports multiple rendering Modes:Immediate, Retained, Compiled-retained

1.5. Justification for Java 3D As A LearningEnvironment for Beginning Students

Although at first glance Java 3D may seem like anAPI most applicable for experienced professionalsdeveloping production quality software, severalaspects make it desirable as an API for beginners.

• Platform independenceStudents usually prefer to work in labs at school,on their home computers, and, even at theirplace of work (for those employed full or part-time). Java and Java 3D's platformindependence greatly simplifies portability, notonly for 3D graphics but also for 2D GUI's.

• CostThe purchase price is zero, and there is nosoftware maintenance cost. This is a significantfactor for schools, as well as for students.

• Programming ParadigmThe fully object-oriented environment isconsistent with training students receive in theirprerequisite courses. There is no need to kludgetogether OO with non-OO procedures.

• High PerformanceStudents are accustomed to high performance --or at least what they perceive to be highperformance. The many PC games and Webapplets that present apparent high performancerendering raise expectations. However, they cancause frustration for students if they are limitedto programming lab examples that appearsimplistic by comparison. A significant aspect ofgraphics education is the motivation provided byits appealing real-time, interactive, visualresults. If that motivation is frustrated by lowperformance, the learning that occurs will bediminished.

5. Experiences in Adapting CourseCurricula to Changes in GraphicsTechnology

We have put our recommendations into practice. InSpring 1999 one co-author taught a revised versionof the Cal Poly undergraduate Introduction toGraphics course using OpenGL and Java 3D APIs.During Winter and Spring 1999 the other co-authortaught three courses, varying in duration from one tothree days, on Introduction to Java 3D forexperienced professionals.

In the current university environment, students areentering graphics courses with less foundationalknowledge than in the past. Although most studentscome to school with much greater exposure tocomputers and graphics, they have less exposure tomathematics, problem solving, and less of an ideawhere they want to focus their careers. Furthermore,the field of graphics has expanded greatly since itsinception, and given the new APIs, computergraphics is no longer the exclusive province ofexperts. The challenge for educators is to provideenough information so students can learn graphicswithout requiring them to becomes masters of aparticular sub-discipline. After all, twelve-year-oldsare producing beautifully ray-traced images withouthaving any concept of how ray tracing works.

A high level graphics API provides educators withthe ability to introduce a broad range of fundamentalconcepts without detouring into the details importantonly to graphics professionals. Despite the higherlevel presentation of concepts, students can quicklylearn how to write useful and compelling graphicapplications. In the process, they are exposed toenough graphics concepts that they can decidewhether to devote more of their student andprofessional career to graphics.In the Cal Poly CSC 471 Introduction to GraphicsSpring 1999 course [9] students used OpenGL andJava 3D APIs, a low level and high level APIrespectively. These APIs allow topics to be coveredquickly in both breadth and depth. Programmingassignments using a high level API (Java 3D) allowstudents to produce quite substantial results within ashort period of time. In one programmingassignment students wrote a Java 3D program todisplay a humanoid-like robot with 16 rotationaljoints and interactive behaviors that allow run-timemodification of the joint angles. This project wascompleted within 3 weeks of their introduction toJava 3D (and, for some students, within 3 weeks oftheir first exposure to Java after having been trainedin C++). At the same time students were learning thefundamental concepts of a scene graph and weresolidifying their knowledge of complex transformoperations, they also learned the details of Java 3Dscene graph implementation.

Experienced computer professionals havesignificantly more exposure to mathematics, problemsolving and logical thinking. However, when theydecide to study graphics, they often revert to“student mode”. Presentation order, thus, remainsimportant. Fundamental material must be introducedin a constructive order rather than deconstructiveorder. Otherwise students become confused andfrustrated. Although experienced professionals maybe able to solve the most complex problems withouthand holding -- outside the classroom -- when theybecome students they require the smallest details tobe completely specified. This is not surprising, sincemuch like university students they also cannotdistinguish between fundamental concepts andunimportant details.

This becomes more of an issue given the breadth anddepth of topics within the graphics field today andthe short time frame of a typical professional shortcourse. Despite this challenge, higher level APIsallow educators to introduce breadth as well as todelve into selected topics in depth within short timeframe courses. By emphasizing fundamental ideas aneducator can start such students in the necessary

direction and leave them with sufficient landmarks toallow the students to explore the details further ontheir own.

6. Summary and Conclusions

Graphics technology has changed vastly in the nearlythree decades since researcher-educators first startedpassing their knowledge on to students. Students andcomputing environments have also changedsignificantly. Course curricula and textbooks havebeen slow to adapt and have not kept up with thesechanges. Proposals for adapting curricula to matchmodern technological requirements have beenpresented here and elsewhere. Personal experiencewith such adaptation in university courses and inindustrial short courses has demonstrated theeffectiveness of using modern, high-level API's andemphasis upon student-centered learning.

References:

[1] Geoffrey A. Moore, Crossing the Chasm,HarperBusiness, 1991.

[2] Alan B. Tucker and Bruce H. Barnes,editors, Computing Curricula 1991: Reportof the ACM/IEEE/CS Curriculum TaskForce, ACM Press/IEEE Computer SocietyPress, 1991.

[3] Edward Angel, Interactive ComputerGraphics: A top-down approach withOpenGL, Addison-Wesley Longman, 1997.

[4] Lewis Hitchner, Steve Cunningham, ScottGrissom, and Rosalee Wolfe, ComputerGraphics: The Introductory Course GrowsUp, Panel session, Proceedings of the 30th

SIGCSE Technical Symposium onComputer Science Education (SIGCSE ’99),New Orleans, LA, USA, March 24-26, 1999.

[5] Lewis E. Hitchner, CSC 455 course webpage, Spring 1998,http://www.csc.calpoly.edu/~hitchner/CSC455.S98

[6] Henry Sowizral, Kevin Rushforth, andMichael Deering, The Java 3D APISpecification, Addison-Wesley Longman,1995.

[7] Henry Sowizral, Kevin Rushforth, andMichael Deering, The Java 3D APISpecificationhttp://java.sun.com/products/java-media/3D/forDevelopers/j3dguide/j3dTOC.doc.html

[8] Java 3D White Paper,http://java.sun.com/marketing/collateral/3d_api.html

[9] Lewis E. Hitchner, CSC 471 course webpage, Spring 1999,http://www.csc.calpoly.edu/~hitchner/CSC471

Computer Graphics in the Scope of Informatics Engineering Education

Manuel Próspero dos SantosComputer Science Department

Faculty of Science and TechnologyNew University of Lisbon

P-2825-144 Caparicaps@di.fct.unl.pt

Abstract

At the Faculty of Science and Technology of the NewUniversity of Lisbon, the education in InformaticsEngineering has been supported by a B.Sc. courseand a M.Sc. course. Both of them include ComputerGraphics lectures, being the correspondingcurricula adapted as much as possible to the otherbasic scientific matters. On the other hand, the toolsand teaching methods have also been accommodatedaccording to the global resources available at theFaculty for those courses, namely the computationalones. The paper presents the methodologies usedtoday for teaching Computer Graphics at the NewUniversity of Lisbon, gives the results of somesucceeded experiments, refers the most relevantrelationships with other subjects and put importantquestions to the future of Computer Graphicseducation in the scope of Informatics Engineering.Keywords: Computer Graphics, Teaching,Education, Informatics Engineering.

1. Introduction

This position paper is about computer graphicseducation at the Faculty of Science and Technologyof the New University of Lisbon (UNL), which wasthe first University conferring a B.Sc. degree inInformatics Engineering in Portugal. The wayteaching has been done and the correspondingcurricula have been chosen can be the object of ageneralized discussion that will eventually lead tosome improvements not only in these courses butalso in similar ones. As it is well known, thecomputer graphics field has a multidisciplinarycharacter and is still evolving nowadays. The impactof new products and different techniques that can beused today, such as the Internet, clearly showsanother orientation a teacher has to give in order tosolve some arising problems.

Although the main focus of the present paper isrelated with the teaching of computer graphics inB.Sc. and M.Sc. courses, the impact of this subjectof study in other areas of Informatics Engineering atthe UNL is also emphasized. Therefore, computergraphics education has been oriented, not only in theperspective of the end-user of a graphical

application, but specially from the point of view ofthe underlying design and implementation. Thus,both the API and the basic programming level of agraphics system are to be covered in the courses.

2. Curricula of Computer Graphics

The next subsections of the paper describe the maincontents of the different courses in computergraphics at the UNL. In addition to the indispensabletextbooks, teacher’s own notes are also available forall these courses. On the other hand, since the impactof the Internet has been enormous, not only this toolis filling the common lack of books and journals inour academic libraries but it is also a way to easilyfind very useful software products.

2.1. The Graduation Course LevelThe existence of distinct modules at different levelsof interest has shown to be very convenient forteaching computer graphics in the graduation coursefor Informatics Engineering, which is five yearslong. In the subsections below one can see thechoices made in the real situation.

2.1.1. FundamentalsObviously, the first of several modules has always tohave a more introductory programme. In the case ofthe B.Sc. course at the UNL, this module is calledComputer Graphics I. It is one semester long and amandatory subject. The topics corresponding to thecovered matters are classified into some relativelydiversified and well-known main areas:

• Computer Graphics definition, history andapplications

• Main algorithms for output primitive rasterconversion and corresponding attributes

• Clipping of lines and polygons

• Geometric primitives in graphical systems

• 2D viewing pipeline: Coordinate Systems andWindow-Viewport transformations

• 2D and 3D Geometric Transformations

• Utility functions for geometric transformationcomposition

• Parallel and Perspective Projections

• Interactive methods for the design of 2D and 3Dcurves, including evaluation and representationtechniques

• Interactive methods for the design of free-formsurfaces, including evaluation and representationtechniques

• Input logical devices and interaction techniques

• Implementation techniques for the Locator e Picklogical devices

• Main control functions and paradigms in bothPHIGS and OpenGL™

• Concepts of hierarchical modeling and editionusing PHIGS, OpenGL™ and VRML

2.1.2. An Advanced CourseBased on the previous knowledge of the studentsabout graphics programming and concepts, a secondmodule – called Computer Graphics II – covers amore restricted domain, although the main textbookis the same in both modules [1]. Due to the recentimportance of 3D graphics, the chosen topics aremainly associated with photorealism and can beclassified into the following groups:

• Introduction to the problem of photorealisticimages

• Integration of geometric projections

• View volumes for arbitrary parallel andperspective projections

• The synthetic camera paradigm

• Normalizing Transformation and PerspectiveTransformation

• 3D clipping in homogeneous coordinates

• Algorithms for Hidden Line and Hidden SurfaceRemoval

• Achromatic color and halftoning techniques

• Fundamentals of Colorimetry

• Color models

• Rules for the use and harmony of colors

• Basic notions of Fotometry

• Illumination models

• Special effects

• Shading methods for polygon meshes

• Texture mapping

• Ray-Tracing

• Radiosity

• Anti-aliasing techniques

The comparison of the curricula of theseundergraduation courses, as they were stated above,may lead to some erroneous conclusions. Forexample, 3D graphics are seemingly postponed tothe advanced course, causing an important lack inthe basic one. However, this is not entirely true. As amatter of fact, what are really the significantcharacteristics of 3D graphics that are not covered bythe topics of the first course? Note that geometricprojections, 3D curves and the design of free-formsurfaces, among others, were already included there.For this reason, we think the missing topics are thefollowing ones: HLHSR algorithms, solid modeling,shading and illumination models. But these issuesare presented to the students during the semesterfrom an end-user point of view, sometimes with thehelp of pictures produced by well-knowncommercial products or extracted from recentmovies. This kind of information is easily found inthe Internet nowadays. If this is so, why don’t wedispense with the second level course? Just because,in that way, the consequent level of students'education would not at all be adequate to anInformatics Engineering graduation. Thus, thespecific topics about 3D appearing explicitly in theadvanced course are treated in a programmer’s pointof view and not only by informing the students aboutthe right application product to quickly produce thefinal result.

2.2. A Post-Graduation CourseOne important goal of the M.Sc. course inInformatics Engineering is the specialization incomputer science of graduated students. Due to theimportance of computer graphics nowadays, thisfield of knowledge could not stay away.

The actual curriculum is based on the followingitems, whose main emphasis will be explained insection 4.:

• The role of Computer Graphics

• Main graphical representation methods

• Elements of Computational Geometry

• Geometric Modeling supporting the constructionof synthetic images

• Solid modeling and representation schemes,namely B-rep, BSP, Octree and CSG

• Reference Models and API standards

• Concepts, paradigms and programming methodsusing PHIGS/PHIGS+, OpenGL™, Java andVRML

• Data Visualization, including isosurfacegeneration, particle systems and volumerendering techniques

3. The Training at the Laboratory

An education in computer graphics is far from beingcomplete without any laboratory component.Students are asked to develop some tasks whoselevel of difficulty depends on the kind of course theyare attending: basic or advanced. The picturesincluded in this section are intended to provide thereader with a pictorial overview of some significanttasks.

Currently, the problems proposed to the students area mixture between algorithmic and project orientedtasks. An effort is made in order to have all mainchapters of the courses present in the final programsdeveloped by the students. Additionally, thetheoretical classes have to be synchronized with theexplanation of the matters that are needed to find themost convenient solution for each problem.

All the pictures shown in the included Figures arescreenshots taken from some programs developed bystudents at the UNL very recently. The programmingactivity in computer graphics has clearly benefitedfrom the general purpose platform given by the Intel-based PC nowadays. For this reason, showingalready a great evolution from the earlyrecommendations stated in [2], at the UNL there isno longer any visible difference between a computergraphics laboratory and another one equipped forunspecified programming training.

The Microsoft’s Visual C++ has been used in aWindows 95 environment [3]. Also the OpenGL™programs were entirely created having the support ofthis platform [4], most of them interfacing thewindow system through the GLUT library [5].

Figure 1 is just one example of window-viewporttransformations and planar geometric projections,with the possibility of an interactive control of theparameters.

Figure 1. Window-Viewport Transforma tionsand Planar Geometric Projections.

Different methods for designing curves areintegrated in a simple editor to be implemented bythe students and illustrated in Figure 2.

The fire-fighting truck depicted in Figure 3 can bemoved forward and backwards under the usercontrol. Additionally, the stairs have two kinds ofindependent movements: an elevation produced by arotation transformation and a variable lengthachieved by shifting the three componentsaccordingly.

Figure 2. Interactive free-form curve editor.

Figure 3. 2D hierarchical modeling withinteractivity.

Figure 4. 3D hierarchical modeling withinteractivity.

The ability of the students to create the underlyinghierarchical model has shown us that not only allimportant concepts can be learned and practised in a2D environment but also with additional advantages:normally the results in a 2D space have an easierinterpretation and are less time consuming fordebugging. The corresponding 3D situation is merelya generalization. This is illustrated in Figure 4.Therefore, the main difficult here is related to the useof illumination and projected shadows. Both Figures3 and 4 resulted from OpenGL™ programs.

Figure 5. Surface rendering.

Figure 6. Illumination models and Shading.

The generation of the pictures shown in the Figures 5and 6 was achieved by writing directly in the framebuffer, thus dispensing with the high-level functionsusually found at the API level of a graphical system.This kind of exception to the use of OpenGL™ (see[6]) gives the students a good understanding of therendering issues and allows them to program anyspecial technique, such as the Phong shading.However, to be a feasible task in a short period oftime, some restrictions had to be made: for instance,the output pipeline was simplified by fixing theviewing parameters.

Figure 7 shows us one single frame from anOpenGL™ animation program using B-splines.

Figure 7. 3D Modeling and Animation.

Finally, it is important to refer that an event-drivenprogramming style is present in all applicationswritten by the students.

4. Problems and Suggested Solutions

In what concerns Computer Graphics I, the topicsdealing exclusively with 2D issues are taking nearly25% of the total time of the actual theoretical classes,being very difficult to make this value lower. Onecannot forget that even what we call 3D graphicsbecome 2D at some stage of the complete outputpipeline. Nevertheless, it is obvious that the referredpercentage would be much more reduced at the costof sacrificing the explanation of several algorithmsand techniques if there were no further coursesavailable, such as Computer Graphics II.

But, in our opinion, that reduction can be achievedby other means, which are compatible with agraduation course having a strong basic kernel ofsubjects. Let us put the following question: shouldwe teach a programming language (e.g. C or C++)within a computer graphics course? The answer isclearly negative, because the knowledge aboutprogramming languages is an attendance pre-requisite for any computer science student. So, whydon’t we try to take a step forward by moving part ofthe traditional computer graphics topics into otherbasic subjects? The main problem here is that aserious compromise must be agreed with theinstructors responsible for those subjects. But theforeseen benefits are great for all parts. Consider, forinstance, the case where the student is learning theconcepts of recursion. What would be the differenceif those studies were complemented with acomparison between recursive and iterativealgorithms for filling regions of pixels?Computationally speaking, this would be a muchmore valuable approach than the traditional

numerical examples such as the factorial function orthe Fibonacci numbers. And there is a lot ofadditional examples from the area of computergraphics that could be useful to other fields ofcomputer science and don’t need long explanationsto the new generation of students.

This solution can lead to a natural integration ofdifferent subjects of the graduation in InformaticsEngineering, but requires the staff responsible for thecorresponding lectures to have also a very goodbackground knowledge about graphics. To achievethis goal, the necessary teacher's education is beingstarted at the UNL in what concerns the preliminaryprogramming language courses. We hope that, withthe functionality present in Delphi, some basiccomputer graphics concepts and techniques will beeasily introduced at the same time students start tolearn Pascal and basic data structures.

There are also several general problems that becamevisible during the past experience in computergraphics education at the post-graduation level inPortugal. The most important is related with thedissimilarity of the background knowledge of thestudents, since they usually come from very differentUniversities in the country. This fact would not beserious if a suitable set of different curricula could beimplemented in order to configure the course to theprofile of each student. However, this methodologyis only feasible when the expected number ofstudents is considerably high. This is not the presentcase at the UNL and no change is foreseen at short orlong term. As a matter of fact, we are talking about aglobal number between 15 and 20 attendants per yearat this M.Sc. course.

As a consequence, the suggested strategy is toelaborate a curriculum where the followingcompromise is assumed. On the one hand, somebasic and general issues in computer graphics haveto be quickly addressed in order to uniformize thebackground knowledge of the students. On the otherhand, the choice of the most convenient topics mustinclude the state of the art techniques and themethodologies that can meet the wishes of a studenthaving already a good understanding of computergraphics fundamentals.

However, no intensive programming is normallyneeded at the curricular part of the M.Sc. courses,since they consist of theoretical lectures and thelaboratory is free to be used by the studentswhenever they want to see the practical aspects ofsome particular matters. Computational geometrytopics and geometric modeling issues are consideredto be a good choice from this point of view, becausethey can be treated at a level as high as we want.Nevertheless, a programming activity is alwayswelcome in Informatics Engineering. Due to itsdeclarative character, VRML was elected as the most

privileged candidate to the main tool for developingskill in graphics programming either at a basic oradvanced level [7, 8].

5. Conclusions

In general, the evolution of educational methodsgreatly results from the analysis of past experiences.But these systems are very complex and the staffresponsible for teaching is not always having theknowledge about all the relevant parameters.Although many teachers conscientiously assume thisfact, that is not the general rule. It thus becomes clearthat the lack of helping mechanisms to solve theproblem also affects those teachers who want toimprove their methods based on the experience fromanother one.

When trying to find an adequate answer to a specificproblem one must obviously consider eachmethodology that proved to succeed in similarsituations before. However, the existence of a smalldifference between them, even not very significantapparently, may lead sometimes to a completelydifferent and unexpected result. It also happens thatthe education and learning processes can be highlyimproved if some details are taken into account. Forinstance, following an observation by Jansen andNieuwenhuizen in [9], the first work given to thestudents attending the laboratory classes in ComputerGraphics I set to be a pre-programmed task insteadof a simple statement about the problem to be solved.And the gratifying result was an extraordinary growin the motivation of the students.

Computer graphics is far from being restricted to 3Dgraphics, no matter their increasing importancetoday. The way the basic undergraduate course isimplemented at the UNL, as described in the paper,does not punish the student who has more difficultiesto picture 3D entities and to realize, for instance,their shape and behavior related with movements,intersections, derivatives, volumes and so on.

The paper addresses other important issues. One ofthem is related with the direct impact the computergraphics education has had in other subjects ofInformatics Engineering courses, specially in thosematters connected to programming languages,algorithms and data structures. But that is not all, aswe can see next. At the UNL there is nocomputational geometry course yet. Thecorresponding most significant issues are onlyreferred in the computer graphics courses. However,we are trying to move part of them into a recentlycreated course on complementary mathematics,named Discrete Mathematics, in the second year ofthe graduation studies. We hope it will be possible toextend this kind of collaboration and integration toother subjects in the near future.

Due to the rapid progress on the computer graphicsfield, the methods and contents of each coursechange every edition at the UNL. This is true evenfor the most basic one. The actual domain ofcomputer graphics is so wide that it is impossible tocompletely embrace it in any course, especially whena small set of classes is provided. Comparing theUNL courses with the HyperGraph project [10], forexample, we can observe that the major set ofuncovered items concerns the hardware descriptionsand not the software issues. In summary, eacheducational institution must choose the most relevanttopics and the way they can be taught in order tocontribute to the specific objectives of the graduationand post-graduation courses to be offered. Lackingfor adequate evaluation methods, this is usually avery subjective and difficult task for the teachers.

In 1999, at the UNL, the introductory course forundergraduate students moved from the fourth to thethird year of the graduation in InformaticsEngineering, reflecting the fact that computergraphics is no longer a complementary subject. As amatter of fact, that course gives such a completepreparation to the students that it can be very usefulto several subjects taught in the last two years of thegraduation time period. One must note that this kindof change is only possible when the requiredextensive amount of background in mathematics andprogramming is guaranteed at the end of the secondyear of the computer science studies.

Acknowledgements

The following students gave their contribution to theFigures of section 3: Hugo Abreu, Paulo Matos,Telmo Cruz, Paulo Santos, Sérgio Estêvão, JoãoFeliciano, João Canhita, Alfredo Pereira, CristinaGarcia, Pedro Valente, Pedro Lopes and JoséDomingos.

References

[1] J.D. Foley, A. van Dam, S.K. Feiner, J.F.Hughes. Computer Graphics — Principlesand Practice. Second Edition in C. Addison-Wesley (1996).

[2] ACM/IEEE-CS Joint Curriculum TaskForce, Computing Curricula 1991.(February 1991).

[3] C. Petzold. Programming Windows. 5th

edition, Microsoft Press (1998).[4] R. Fosner, OpenGL Programming for

Windows 95 and Windows NT. Addison-Wesley (1996).

[5] J. Neider, T. Davis, M. Woo (OpenGLARB). OpenGL Programming Guide.Second Edition, Version 1.1. Addison-

Wesley (1997).[6] R. Wolfe. OpenGL: Agent of Change or

Sign of the Times?. Computer Graphics. 29-31 (November 1998).

[7] A.L. Ames, D.R. Nadeau, J.L. Moreland,The VRML 2.0 Sourcebook. 2nd edition. JohnWilley & Sons Inc (1997).

[8] B. Roehl, J. Couch, C. Reed-Ballreich, T.Rohaly, G. Brown. Late Night VRML 2.0with Java. Ziff-Davis Press (1997).

[9] F.W. Jansen, P.R. Nieuwenhuizen.Computer Graphics Education at Delft

University of Technology. Proceedings ofthe Eurographics Workshop on Graphicsand Visualization Education, Oslo (1994).

[10] G. Scott Owen (project director).HyperGraph — Teaching ComputerGraphics: A project of the ACMSIGGRAPH Education Committee, theHypermedia and Visualization Laboratory,Georgia State University, and the NationalScience Foundation.http://www.education.siggraph.org

A Course in Topology-based Geometric Modeling

Pascal Lienhardt, Laurent Fuchs and Yves BertrandIRCOM-SIC,

Poitiers, France{lienhardt, fuchs, bertrand}@sic.sp2mi.univ-poitiers.fr

Abstract

The content of a course in Topology-based GeometricModeling is presented. The key ideas are based uponsimplicial notions issued from combinatorial topology.The well adequacy of these notions to geometricpatches (such as Bézier patches) is pointed out.Topology and geometric notions are simultaneouslypresented in order to show their complementarity.Keywords: Geometric modeling, combinatorialtopology, subdivisions, geometric patches.

1. Introduction

The goals and the content of a course in Topology-based Geometric Modeling are described in this paper.Since several years, it is addressed to ComputerScience students (Master Degree, ninth and tenthsemester) who have acquired basic knowledge incomputer graphics, but who did not thoroughlyapproach Geometric Modeling. Other coursesspecialized in other fields of Geometric Modeling cansupplement this course, e.g. C.S.G. and implicitsurfaces.

The purpose is the acquisition of solid basic concepts:so, we insist here about the fundamental concepts inmathematics and computer science, and mainly about :

� (combinatorial) topological notions related tosubdivisions of geometric spaces,

� parametric curves and surfaces, and moregenerally curved shapes,

� the links that can be made between these twonotions, especially combinatorial ones.

The design of data structures is considered as well asthat of basic construction operations and algorithms forcomputing topological properties.

The course contains three main parts :

� simplicial notions, mainly simplicial sets andtriangular patches ;

� simploidal notions, where cells are cartesianproducts of simplices ;

� cellular notions, where cellular structures arededuced from “numbered” simplicial sets andembedding is based upon parametric trimmedpatches.

We insist here about the complementarity between thetopological aspects (subdivision, etc) and theembedding ones (curves and surfaces). Highlightingthe common combinatorial concepts facilitates this.Hence a global and homogeneous outline of the field isgiven.

2. Course content

2.1. Introduction (2h)

Basic concepts are introduced: subdivision ofgeometric space (i.e. partition into cells of variousdimensions on which boundary relations are defined),various "classes" of subdivisions (simplicial or cellularones, complexes, manifolds, ...), homeomorphism, etc.

This initiation is carried out by the means of aconstructive approach of the classification of surfacesubdivisions into topological surfaces, as it can befound in [8]: the basic objects are polygons(topological discs) and surface subdivisions are builtby identification of edges. The classification ofsurfaces (and the related characteristics: number ofboundaries, orientability factor, genus) is approachedby distinguishing the various cases of identification(the proof that any subdivision belongs to a uniquetopological surface is not tackled).

This part can by supplemented by a practical sessionduring which the students, provided with paper,scissors and tape, build and classify paper surfaces.

2.2. Simplicial : Fundamentals (6h).

A first part shows how to deduce combinatorial notionsfrom geometrical ones. We define successively:

� simplicial complexes, defined as convex hulls oflinearly independent points,

� abstract simplicial complexes, where a simplex isa set of abstract vertices ; the geometric realizationof an abstract simplicial complex is a simplicialcomplex (vertices are associated with points, ...),

� semi-simplicial sets [12], where a simplex is anabstract object and boundary operators associateits boundary simplices with the simplex (cf. thefollowing definition and Figure 1). The geometricrealization of a semi-simplicial set is a CW-complex.

Definition [12] : A d -dimensional semi-

simplicial set K , is a union �d

l lKK0=

= of sets

lK of abstract simplices of dimension dl ≤ ,

together with boundary operators i∂ such that:

1: −→∂ lli KK with 1≥l and li ≤≤0 ,

1−∂∂=∂∂ ikki σσ for ki > and K∈σ .

0 0

1 1

0

2

1 0

1

1 0

Figure 1. 1- and 2-dimensional simplices, andtheir boundaries.

Basic additional notions and construction operationsare defined (e.g. geometric realization, isomorphism,simplex creation, identification) and traversalalgorithms are described in order to computetopological properties (e.g. connectivity, boundary,star).

In a second part, the embedding notion is introduced inorder to define the object shape. First, we show that asubclass of semi-simplicial sets is equivalent toabstract simplicial complexes ; so, they can beembedded as simplicial complexes, by associatingpoints with 0 -simplices.

Embedding is then generalized by taking curved shapesinto account (i.e. curves, triangular surfaces, tetrahedralvolumes, etc). These shapes are introduced by the"Bézier simplex'' notion and described using the deCasteljau algorithm in any dimension. Bernsteinpolynomials are defined. Joining and continuityproblems are taken into consideration. Figure 2 givesus an example of a semi-simplicial set.

Combinatorial links between an abstract simplex and aBezier simplex are pointed out. Mainly it is possible toextract a semi-simplicial set structure from the controlpoint structure (cf. Figure 3) : it is thus possible toassociate subsets of control points with abstractsimplices, and this can be used in order to design datastructures.

Figure 2. Example of semi-simplicial set, upthe topological structure, down the embedding.

003

111

120210

102

201

300 030

012

021

30

30 21 03

30

03

21

12

03

12

21

12

Figure 3. Semi-simplicial set structure andcontrol point structure.

The last part is devoted to geometric operations (e.g.splitting, extrusion and cone). They are compositionsof operations acting on semi-simplicial sets and onBézier simplices.

Implementation is also addressed from abstract datatypes to concrete data structures using records andpointers. A basic concrete data structure can be directlydeduced from the combinatorial definition of n -dimensional semi-simplicial sets.

struct cpl {(I+1)_SIMPLEX * s ;int b_op ;struct cpl * next, * prev ;

} * CPL ;

struct i_simplex {struct simplex * delta[i+1] ;CPL ante ;

} I_SIMPLEX ;

The record (written in pseudo-code) I_IMPLEX

represents an i -simplex : the first field (delta )implements the boundary operators ; the second fieldimplements the "inverse" operators : a doubly-linkedlist contains the simplices which have the consideredsimplex as image by a boundary operator.

From an abstract data type point of view, the simplicialsets are defined by the operations working on the datastructure. Basic operations such as

set_border : nat simplex simplex simplexSet -> simplexSet

border : simplex nat -> simplex

which respectively define boundary operators andapply boundary operator to a simplex are defined byequations. Then fundamental properties are expressed,e.g.

border(k, border(i, s)) = border(i-1, border(k, s)) if i > k

Similar implementation mechanisms are used for theother simploidal and cellular notions presented in thenext sections.

2.3. Simplicial (4h).

This part has three complementary goals:

� Extend semi-simplicial sets by defining simplicialsets [12] ; thus cartesian product can be defined.This operation is a generalization of the extrusionone; it is the combinatorial basis for thecomputation of Minkowski sums. Moreover,

notions presented in section 2.4 are based uponcartesian product.

� Define basic mechanisms which make it possibleto deduce optimized structures for some classes ofsimplicial sets (e.g. homogeneous, regular, quasi-manifolds), and related operations (i.e. constrainedoperations that are internal to sub-classes). Usingthese mechanisms, relations can be establishedwith other structures defined in literature, and withcellular structures (see section 2.5) [4, 9, 11].

� Notions related to curved shapes are abstractedand generalized by introducing the generalsimplicial algorithms [6, 7]. This provides auniform approach for the parametric curves,surfaces, volumes, etc.

2.4. Simploidal (4h).

Previous simplicial notions are generalized by definingstructures that can represent subdivisions in which cellsare cartesian products of simplices.

First, cubic sets are defined [13] : a cube is thecartesian product of 1-dimensional simplices.

Definition: A d -dimensional cubic set K is the

union �d

l lKK0=

= of sets lK , which elements

are abstract cubes of dimension dl ≤ , together

with boundary operators ij∂ such that:

1: −→∂ ll

i

j KK with 1≥l , 1,0=j and li ≤ ,1−∂∂=∂∂ i

j

k

l

k

l

i

j σσ for ki > and K∈σ .

Combinatorial links between Bézier patches (tensorproduct of Bézier curves) and abstract cubes arepointed out, leading to the embedding of cubic setsusing cubic shapes.

Abstract simplices and abstract cubes are generalizedby abstract simploids (the corresponding geometricnotion is the simploid notion [3]). They are cartesianproducts of abstract simplices. An abstract simploid isfully defined by the k -tuple ),,( 1 kdd � of the

dimensions of the k simplices involved in the product(cf. Figure 4).

*

*

**

*

(2)(1)

(1,1)

(2,1)

(1,1,1)

(2,2)

Figure 4. Example of simploids. The cartesianproduct of at least two simplices is graphicallyrepresented.

Then, simploidal sets are defined by (cf. Figure 5) :

Definition [7] : A d -dimensional simploidal set

K is the union �d

l lKK0=

= of sets lK , of

abstract simploids of dimension dl ≤ , together

with boundary operators ij∂ such that1:

∂∂>∂∂

=∂∂

>>∂∂=∂∂ >−

=∂

−

−

else ),,,,(

1 if ),,,,(),,,,(

1 and with),,(),,(

else ),ˆ,(

1 if ),1,(),,(

1

1

j

k

i

lji

i

j

k

i

ljii

l

j

kji

i

i

k

i

li

i

l

i

ki

i

iii

ji

dd

ddddd

dlkdd

d

ddd

���

���

���

����

��

��

��

Products of general simplicial algorithms [6],generalizing cubic shapes, are also defined, and usedfor embedding simploidal sets. Once again, thecomplementarity of the embedding and the topologicalaspects are pointed out and used. For instance for thedesign of data structures, we also explain how commoncombinatorial notions simplify the implementation.Simplicial operations (and implementation aspects) areextended to the simploidal framework. Lastly, openresearch problems are highlighted.

1 The hat notation means that the element is removed fromthe tuple.

Figure 5. Example of simploidal set.

2.5. Cellular Structures (8h)

A first part shows how to deduce a first cellularstructure from a subclass of simplicial sets [11].Simplicial sets are structured into cells by associatingnumbers with 0 -simplices (this is quite close to theclassical notion of barycentric triangulation : numbersassociated with the vertices of an i-simplex are {0, 1,..., i}). Then, we define cellular quasi-manifolds as asub-class of such "numbered" simplicial sets, and theequivalent notion of combinatorial generalized map isdeduced by applying conversion mechanisms presentedin section 2.2 (cf. Figure 6).

2

1

1

1

1

0

00

0

0

21 1

����

��������

����

����

����

����

�� ������

�� ������

��

����

������������

����

����

���� ������

Figure 6. Numbered simplicial set andgeneralized map. A dart corresponds to a 2-dimensional simplex {0,1,2}. Application αi

joins simplices which share a 1-dimensionalsimplex {0,1,2}-{i}.

Definition: A d -dimensional generalized map (orG-map) is a set D of darts together withapplications (cf. Figures 7,8,9) :

.1 with

, with

,0 with:2

dlkbb

Dbbb

dkDD

kllk

k

k

≤<+=∈=

≤≤→

αααααα

Figure 7. Example of a generalized map.

Construction operations are defined for handlinggeneralized maps, and also algorithms for computingtopological properties [1, 11] (e.g. connectivity,orientability; characteristics can be computed on 2 -dimensional generalized maps providing theirclassification in the meaning of section 2.1).

Figure 8. A subdivision of a Klein bottle whosetopology is represented by a G-map ; itstopological characteristics can be computed bysimple traversal operations (i.e. number ofboundaries = 0, orientability factor = 2, genus= 0).

A second part shows how to deduce cellular structuresfor other classes of "cellular objects" [4] (fromcombinatorial map chains for representing cellularcomplexes, to oriented combinatorial maps forrepresenting oriented quasi-manifolds without

boundaries). These conversions are obtained mainly byapplying mechanisms presented in section 2.2.Relations with equivalent structures are alsohighlighted (e.g. winged-edge, quad-edge, radial-edge,facet-edge, cell-tuple structures, incidence graphs, etc.See [10] for instance).

Figure 9. Complex objects can be modeledwith G-maps

The last part is devoted to embedding. First linearembedding is described. After, parametric trimmedshapes are introduced. They generalize the notion oftrimmed patches [5]. Hence, the link between thetopological structure and embedding becomes moresophisticated: the space of parameters must betopologically structured.

Again, several open research problems are pointed out.

3. Remarks and Conclusion

Our experience shows that a practical work is useful :all notions can be perceived as theoretical ones andintrinsically interesting ; a practical work, for instancethe conception of the kernel of a geometric modeler,shows the usefulness of these notions for the design ofdata structures and algorithms.

Note that sections 2.1, 2.2 and/or 2.5 can be extractedand taught in other curricula. Parts 2.1 and 2.2 havebeen taught to undergraduate Computer Sciencestudents : so, they can experiment simplicial structureswhich are graph generalizations. It is also possible tostudy only cellular structures, by removing sections2.2, 2.3, 2.4. It is thus necessary to give an intuitiveintroduction, e.g. "how to deduce a generalized mapfrom a cellular manifold?” (for instance usingbarycentric triangulation, splitting cell boundariesand/or cell-tuples [2]). But some notions can not becompletely explained (e.g. definition of cartesian

product of cellular structures). At last, the course canbe improved by studying other important notions, e.g.algorithms for computing homology groups, accordingto the background of students and the pedagogicalcontext.

To sum up, this course is based upon simplicial notions(topology: simplicial sets; embedding: curved shapes)and two mechanisms:

� cartesian product providing spaces with simploidalcells,

� numbering 0 -simplices, providing cellularstructures embedded with parametric trimmedshapes.

This approach presents mathematical andcomputational aspects. Fundamentals are explained.Indications are given about higher-level aspects, someof them still being subjects of active research.

Topology and embedding are simultaneously presentedin order to show their complementarity and to highlighttheir common combinatorial notions, which areimportant for the conception of data structures andalgorithms.

All topological notions are based upon simplicial onesissued from combinatorial topology. This discussionthread is useful for the study of deduced structures, andsimplifies the explanations about topological andembedding concepts, operations and algorithms.Moreover, this approach leads coherently to manyother works in geometric modeling, and mainly inBoundary Representation.

References

[1] Y. Bertrand, J.-F. Dufourd, J. Françon andP. Lienhardt. Algebraic Specification andDevelopment In Geometric Modeling. Conf.Tapsoft’93, LNCS n°668, pages 75-89,Springer-Verlag, April 1993.

[2] E. Brisson. Representing GeometricStructures in d-dimensions : Topology andOrder. Discrete and ComputationalGeometry, 9, pages 387-426, 1993.

[3] W. Dahmen and C. A. Micchelli. On theLinear independence of Multi-Variate B-splines I. Triangulation of Simploids. SIAMJ. Numer. Anal. 19, pages 993-1012, 1982.

[4] H. Elter and P. Lienhardt. CellularComplexes as Structured Semi-simplicialsets. International Journal of ShapeModeling, 1(2) pages 191-217, 1994.

[5] G. Farin. Curves and Surfaces for Computer

Aided Geometric Design: A Practical Guide.Academic Press, San Diego, 1988.

[6] L. Fuchs. Spécification formelle desmodèles de courbes et de surfaces de formeslibres. PhD Thesis, Université Louis Pasteur,Strasbourg, France, 1997.

[7] L. Fuchs and P. Lienhardt, TopologicalStructures and Free-Form Spaces. Journéesde Géométrie Algorithmique, Barcelone,Spain, September 1997.

[8] H.-B. Griffiths. Surfaces. CambridgeUniversity Press, 2nd edition, 1981.

[9] V. Lang and P. Lienhardt. GeometricModeling with Simplicial Sets. Proc. ofC.G.I.'96, Pohang, Korea, 1996.

[10] P. Lienhardt. Topological Models forBoundary Representation : a Comparisonwith N-Dimensional Generalized Maps.Computer-Aided Design, 23(1), pages 59-82,1991.

[11] P. Lienhardt, N-Dimensional GeneralizedCombinatorial Maps and Cellular Quasi-Manifolds. International Journal ofComputational Geometry and Applications,4(3), pages 275-324, 1994.

[12] J. P. May. Simplicial Objects in AlgebraicTopology. Van Nostrand, Princeton, 1967.

[13] J. P. Serre. Homologie singulière desespaces fibrés. Ann. of Mathematics, 54,pages 425-505, 1951.

Re-Inventing the Introductory Computer Graphics Course:Providing Tools for a Wider Audience

Steve CunninghamComputer Science

California State University StanislausTurlock, CA 95382 USA

rsc@csustan.edu

Abstract

In ACM/IEEE Curriculum 91 [TUC 91] and inearlier curriculum statements on computer science,the computer graphics course in computer sciencefocused on fundamental algorithms and techniquesfor creating images and animations. Forthcomingcurriculum statements are going to find a need toexpand the role of courses such as computergraphics in the undergraduate computer sciencecurriculum [CUN 98]. In the last few years theapproach to teaching this course has changed totake advantage of more powerful graphics tools.By using these tools, however, we are also able tomake the graphics class accessible to students fromother areas. This paper describes a set ofpresentations, examples, and projects that supportsthe introductory computer graphics course as a toolfor the student in the sciences, mathematics, orengineering as well as for the computer sciencestudent.

1. Introduction

Computer graphics is an important tool for theworking scientist, mathematician, and engineer,and has become a commodity item in thesciences just as it has in so many other areas oflife. Its value is expanding as the scientificculture embraces scientific visualization and aspowerful computers become more and morecommonly available. However, the traditionalcomputer graphics course focuses on fundamentalgraphical algorithms and processes, not high-level graphics programming. This is oriented tothe computer science student who wants topursue advanced work in the field, not the moregeneral student in science or engineering.

This offers computer science anopportunity to serve this wider audience bycreating an introductory computer graphicscourse that has dual goals. One goal is tointroduce computer science students to basicgeometric processes and to graphicsprogramming, and the other is to give studentsin science, mathematics, and engineering a basic

background in thinking and programming in 3Dgeometric terms.

2. The Course Philosophy andOutline

One fundamental goal of the revised introductorycomputer graphics course sets it apart from thetraditional course. This goal is to create asound course for science, mathematics, andengineering students (hereafter abbreviated as“science”) that will give them the graphics andvisualization tools that they need for their future[BRO 90]. In doing this, we must avoidcreating a course that poses significantprogramming challenges for students, becausemany students from these fields will have lessprogramming experience than a computer sciencestudent would. Thus the course must provide asound graphics experience for a student who hashad only a two-semester sequence inprogramming.

At the same time, another course goalis to provide computer science students with anintroduction to the field of computer graphics.Thus the course must be useful to the generalcomputer science student and provide him or hersound skills in graphics programming, becausemany applications will need to add graphicalcontent in the future.

All the work in the course is doneentirely in 3D, with no reference at all to 2Dgrapics except perhaps in a few specificexamples. This supports both goals for thecourse and is consistent with current directionsin API development and in graphicsapplications. The heterogeneous nature of theclass does offer a challenge, but the expectedbenefits should outweigh this problem.Certainly computer science students shouldbenefit from working with others outside ourfield, and science students should benefit fromseeing a computer science approach that supportstheir fields.

The course outline is straightforward.It begins with an introduction to thefundamental notions of geometry as used by

OpenGL, and with simple examples of OpenGLprograms to help students see how to build theirfirst programming projects. It goes on to coverfundamentals of polygon-based modeling,including the use of instancing transformations;to discuss rendering including projections,viewing, lighting, and shading; and to workwith event-handling to provide user controls andanimation. It ends with some of the features thatstudents rarely actually use in a traditionalfundamentals course, such as alpha blending,texture maps, and fog.

This course is largely possible becauseof advances in low-cost and high-functiongraphics systems. Computers with thesecapabilities are now available for even budget-strapped universities. Graphics programmingAPIs such as OpenGL and Java3D are readilyuseable by students who have a programmingbackground but are not computing specialists.Further tools such as the Fahrenheit API beingdeveloped by SGI and Microsoft [FAR 98] willalso soon be widely available.

3. The Presentation Sequence andExamples

In order to introduce students to the subject, webegin with geometric and programmingfundamentals and proceed to more sophisticatedtopics in modeling and rendering. The listbelow summarizes the contents, but it shouldnot be taken as an indication that we proceed ina linear way through this material. Instead wedo a little modeling, then a little rendering, thencome back to more modeling, then morerendering, and so on until we have covered allthis content in one semester.

In order to be valuable for the broadaudience we noted above, the course materialneeds to be backed up with a collection ofexamples and projects that come from an equallybroad set of areas. We need some examples andprojects from mathematics, some from physics,some from chemistry, and some from variousbranches of engineering. Creating a set ofmaterials with this broad coverage will takesome time, and the author has begun to work onsuch a project. These materials will bedeveloped and tested in collaboration with otherfaculty having many different kinds of experienceand with students from other disciplines. Thegoal is to have a draft set of materials availableby mid-2000 and a completed set a year fromthen. The materials will be distributed on theWeb and will continue to evolve as ourunderstanding of the needs of science andengineering students grows.

An outline of the overall set of topics covered is:

I. Introduction to geometry and to graphicsprogramming

a. Coordinate systems and geometryfundamentals in 3-space.Coordinates, normals, vectors, dotand cross products, polygons, etc.

b. What is an OpenGL program?Example or two of some simplecode in template programs tointroduce the idea of using theOpenGL system.

II. Modeling

a. Modeling primitives. Points,lines, triangles, quads, trianglestrips, triangle fans, quad strips,etc. Using built-in GLUTprimitives; modeling using them.Frames as transformed modelingcoordinate systems.

b. Modeling objects usingtransformations. Creating items fora scene by starting with primitiveobjects and using scaling, rotation,and translation to shape them andplace them in a scene. Handlingtransformation stacks. How thestack of modeling transformationsis used to handle several objects.LIFO behavior of modelingtransformations and the distinctionbetween function composition andstack behavior.

c. Evaluators and Bézier splines.Defining control points; specifyingevaluators; using evaluator togenerate points in the line orsurface.

d. Scene graphs. Concept, examples,applications

III. Rendering

a. Projections and viewing.Orthogonal and perspectiveprojections, standard view volume.Setting the viewpoint; relation toboth kinds of projections

b. Hidden surfaces. Enabling andspecifying; what effects do you get

c. Lighting and materials. Lights in ascene; ambient, diffuse, andspecular lighting. More than onelight. Colored lights. RGB andother color models. How materialsare specified; application to objectsin a scene.

d. Shading. Specifying shading; useof normal vectors. The differencebetween flat and Gouraud shading.

IV. Interaction

a. Event and event handling concepts.Event queues, event types, conceptof callbacks, registering callbacks.Keyboard, special keys, menu,mouse, idle

V. Additional features

a. Color blending (alpha channel).Specifying, effects of blends

b. Fog. Enabling; effect of fog color;effect of density; specifying frontand back

c. Textures. Creating texture maps,specifying maps, 2D and 3Dtexture maps

d. Text as a graphic element. Labels,in-space menus, billboards.

As each part of this material is presented to theclass, the presentation first focuses on what thefeature means in creating or interacting with animage by presenting demonstrations of thefeature’s effect on an image and pointing outhow the feature is invoked in code by showingcode examples and experimenting withparameters. When the student has some idea ofwhat the feature is, the presentation will go onto give some background on what the featuremeans in terms of the actual graphics systemwork, and on occasion to include a briefdescription of the algorithms involved. To putthese in balance, if we consider the OpenGLProgrammers Guide [WOO 99] on one hand anda standard textbook that uses OpenGL on theother [ANG 97], the presentation is in based80% on the programmers guide and 20% on thetextbook.

4. The Projects

One key to making this course effective forstudents in science, mathematics, andengineering is creating a good set of projects thatare meaningful to the student. We cannot use asingle set of projects or projects that ask astudent to create good-looking but meaninglessimages. The author is working on a set ofprojects that will fit this orientation for thecourse, and hopes to have a set ready todistribute in mid-2000. Some initial projectsthat were used in the first instance of the courseincluded:

a) 3D histograms that can be rotatedaround one axis; only ambient light

b) physical object with one light source,where both object and light source canbe rotated

c) pseudo-fractal landscape with semi-transparent water and both snow andtree lines

d) N-body gravity problem with eachbody showing a short-term trace of itsrecent location, with time-stepanimation

e) a Bézier surface with selectable andmoveable control points

All of these projects included user-controlledmotions or some form of animation, and someexamples of student work on these projects isshown below. Because we had not yet done anypromotion of this course to students outsidecomputer science, we did not have any studentinput into interesting project topics for non-computer science students. This will be oneapproach to developing such projects in futurecourse offerings.

The language used for projects, either Cor C++, is not an issue in the course or projects.It is possible to call any of the OpenGLfunctions through either language. Thusstudents can choose the language they are mostfamiliar with and use whatever level of thelanguage one can handle. In the future we expectto consider offering the course with Java3D orperhaps Fahrenheit, depending on theavailability of the systems and the amount ofprogramming background needed to use them.A relatively modest independence on theprogramming language is a very important partof making the course work for non-computerscience students and cannot be overemphasized:students can create interesting and useful imageswithout using sophisticated programmingtechniques.

5. The Sequel to This Course

At the end of the introductory computer graphicscourse, students have experience in thinkinggeometrically and in creating programs thatexpress that geometric thinking and use somemodest animation and interaction. They haveseen a very modest description of how somegraphics algorithms are used to create images,but have done no work directly with thosealgorithms. This gives my computer sciencestudents an excellent background for tackling asecond graphics course containing the traditionalfundamental algorithms content.

In the second course, then, we includefamiliar fundamental techniques. These includeinterpolation material such as scan-converting

polygons, shading models, Z-buffering, bumpand environment mapping, as well as techniquessuch as antialiasing and ray tracing. We havefound that straightforward modifications to theBresenham algorithm [CUN 99] provide ageneral and straightforward approach to thewhole range of interpolation algorithms onpolygons, and between this approach and theexperience students get from the first course,they are able to develop a good understanding ofadvanced graphics techniques.

6. Results so Far

The author has taught this course and its sequelonly once, so we cannot give a good evaluationof this work. My computer science students areenthusiastic about this approach to the courseand several took the fundamental graphicsprinciples course that was taught as a secondcourse in sequence. The results of the secondcourse showed that students did learn many keyconcepts in the subject from the first course, andwe were able to move quite quickly throughmaterial that had been fairly painful.

The value of the traditional computergraphics principles course does not seem to bereduced by beginning computer graphics studieswith this new kind of course. Instead the newfirst course makes a very good introduction tothe principles course by providing students witha background in realizing images from geometricconcepts. Starting with the programming courseallows a traditional principles course to movemore quickly and to address more advancedconcepts than a course that assumes no graphicsbackground. The combination of a first coursebased on graphics API programming and afundamental principles course seems to be anexcellent way to introduce computer graphicscontent for computer science students. Moreadvanced courses can be developed as neededwhen additional topics need to be covered.

In order to understand the level of thisgraphics programming approach, it is useful tosee the kind of work done by students in thenew first course. Below are some images fromthe last three of the course’s five projects. Eachproject also involves animation and/or userinteraction in addition to the image synthesis.

Figure 1. pseudo-fractal landscapewith alpha blending in water

Figure 2. n-body problem withfade-out trails on bodies

Figure 3. Bézier spline surfacewith control points shown;

one is selected for movement

In the first example, students use keyboardcallbacks to move around the landscape and seeit from different viewpoints, and use mousecallbacks to work with a menu to change variousfeatures of the landscape. In the second example,the bodies move in real time driven by the idlecallback, using simulated gravitational forcesbetween the multiple bodies. In the thirdexample, students move around the shape withkeyboard control and can select and manipulate a

control point with the mouse, as shown; theshape of the surface changes as the point ismoved with keyboard control.

7. Summary

One of the challenges of computer science is tofind ways to serve our universities and stillmaintain the quality of our courses andprograms. Because computer graphics is souseful to so many different fields, it is a naturalcourse to take on a service role, but because thecourse has traditionally made significantmathematical and programming demands on itsstudents, it has rarely been used in this way.

The approach outlined in this noteoffers us a way to provide a course that benefitsboth the computer science student and thestudent from mathematics, science, orengineering. It is accessible to faculty, is usefulto computer science students, opens the door tostudies of more advanced computer graphicstopics, and is a valuable service to otherstudents whose professional lives will requirecontinual use of computer graphics.

And — let’s say this only here at theend, and let’s say it quietly, though it makes areal difference to the nature of the course — thecourse is a lot of fun to teach and students reallyenjoy the high-quality work they produce in thecourse.

References

[ANG 97] Angel, Ed, Interactive ComputerGraphics: A Top-Down Approachwith OpenGL, Addison-Wesley,1997

[BRO 90] Brown, Judith R., SteveCunningham, and Mike McGrath,“Visualization in Science andEngineering Education,” in IEEETutorial: Scientific Visualization,Nielson, Gregory M. and BruceShriver, eds., IEEE ComputerSociety, 1990

[CUN 98] Cunningham, Steve, “Outside theBox — The Changing Shape of theComputing World,” invitededitorial, SIGCSE Bulletin 30(4),December 1998, 4a-7a

[CUN 99] Cunningham, Steve, “TheUbiquitous Bresenham Algorithmand Its Applications Across ManyGraphics Processes,” in preparation

[FAR 98] SGI Fahrenheit Web site,http://www.sgi.com/fahrenheit/

[NAD 98] Nadeau, David R. and HenrySowritzal, “Introduction toProgramming with Java3d,”Eurographics 98 tutorial, alsopresented at SIGGRAPH 98

[TUC 91] Tucker, Allen B. and Bruce H.Barnes, eds., Computing Curricula1991: Report of the ACM/IEEE/CSCurriculum Task Force, ACMPress/IEEE Computer Society Press,1991

[WOO 98] Woo, Mason and Dave Schreiner, AVisual Introduction to OpenGLProgramming, SIGGRAPH 98course notes, 1998

[WOO 99] Woo, Neider, and Davis, OpenGLProgramming Guide, third edition,Addison-Wesley, 1999

A Mathematica Package for CAGD and Computer Graphics

Andrés Iglesias1, Flabio Gutiérrez1,2 and Akemi Gálvez1

1Departament of Applied Mathematics and Computational SciencesUniversity of Cantabria, Santander, Spain

2Departament of Mathematics, Sciences FacultyNational University of Piura, Peru

iglesias@ccaix3.unican.es, flabio@tallan.unp.edu.pe, uc8031@cclx1.unican.es

Abstract

This paper introduces a Mathematica package forlearning and teaching Computer-Aided GeometricDesign (CAGD). The package incorporatescommands for the treatment of the most usual curvesand surfaces in CAGD (Bézier, B-spline, rationals,etc.). The powerful symbolic and graphicalMathematica capabilities, the functional and patternrecognition programming and the Mathematicavisualization environment have been extensivelyapplied to get a user-friendly, didactic and powerfultool for education, with applications in areas asCAGD and Computer Graphics. Some illustrativeexamples showing the performance of the packageare also given.Keywords: Symbolic programming, visualization,computer graphics, CAGD.

1. Introduction

Computer graphics and geometric modelingconstitute fundamental parts of the engineeringdesign process. Geometric modeling methods areapplied to construct a precise mathematicaldescription of the shape of a real object and itsefficient computer representation. On the other hand,computer graphics capture the visual and quantitativeaspects of this object. They are also essential topicsin the engineering education process (see [1]).

1. In the last few years, engineering teachers havewitnessed an extraordinary change in the way ofteaching and developing students' skills for theirprofessional future. The extraordinary advances inhardware and software have made possible theappearance of a new generation of scientificSymbolic Computation Programs (SCPs), asMathematica or Maple. In comparison with thetraditional program-ming languages, (Fortran,Pascal, C, etc.) the SCPs incorporate a great numberof symbolic, graphical and programming facilities.Today, teachers can take advantage of thesegraphical capabilities when introducing students tocomputer graphics and/or visualizing the geometry

of the objects under analysis. In addition of this, aswe show below, SCP symbolic tools allow to get notonly a graphical output but also a mathematicalexpression for curves and surfaces.

Mathematica [2] has emerged as the most popularand widely used symbolic computation program. Inthis paper, we present a Mathematica package,CAGD.m, for learning and teaching CAGD andComputer Graphics. The package incorporates manycommands for the treatment of the most usual curvesand surfaces in CAGD (Bézier, B-spline, rationals,etc...). Some classical geometric operations withinteresting applications to real-world problems havealso been performed. In the following paragraphs,we describe the main features of the package andshow some interesting and illustrative examples ofits application.

2. Description of the Package

Tha package CAGD.m to be introduced in this paperhas been implemented in the Mathematicaprogramming language. We think that Mathematicais an ideal environment for developing programs forcomputer graphics, because of its powerfulprogramming language (supporting procedural,functional, based-on rules, based-on patternrecognition and object-oriented programming), andbecause it is easy to understand, it reproduces manyC language structures, it has several remarkablegraphical capabilities and, mainly, it works in asymbolic way. This important feature will be usedlater to obtain the symbolic expression of somecurves and surfaces.

2.1. How Does the Package Works?Perhaps the best way to understand how the packagehas been built is to take a look to some of theroutines it incorporates. For example, letP={P0,P1,...,Pn} be a set of points in Rd, d=2,3 (Rmeaning the set of real numbers). As already know(see [3]) the Bézier curve associated with the set P isdefined by:

where Bi,n(t) represents the Bernstein polynomials,which are given by:

n being thepolynomial degree. The points Pi are called controlpoints or Bézier points of the curve. Theseexpressions can be easily translated to Mathematica.For example, Equation 2 becomes:

Bernstein[i_,n_,v_]:= Binomial[n,i]*(1-v)^(n-i)*v^i

Table 1. Code for the Bernstein polynomials.

while Equation 1 takes the simple form:

BezierCurve[pt:{{_,_}..}|{{_,_,_}..},v_]:= Module[{n=Length[pts]-1}, Simplify[ Table[Bernstein[i,n,v],{i,0,n}].pt ] ];

Table 2. Code for the BézierCurve command.

Let us analyze the previous code: in line 1 weintroduce the variables, namely pt indicating the(two- or three-dimensional) points, and v thevariable of the curve. After obtaining the curvedegree (line 2), we translate Eq. 1 to Mathematica.Thus, line 4 corresponds to the scalar product of twovectors: the table with the Bernstein functions andthe vector of points. Finally, the result is simplified.

From the previous codes, it becomes clear that:

1. Mathematical expressions (as Eqs. 1 and 2) canbe easily implemented in Mathematica; in mostcases, this process consists of a simple translationof these expressions to its programminglanguage.

2. Note also that the previous code allows us towork in a symbolic way. For example, afterloading the package:

In[1]: <<CAGD.m

we can obtain the following mathematicalexpressions for the Bernstein polynomial and theBézier curve:

In[2]: Bernstein[1,3,t]Out[2]: 3(1-t)2 t

Fig. 1 shows the Bernstein functions of degree 4:

Figure 1. Bernstein polynomials for n=4.

In[3]: pts={{1,1},{2,4},{3,0}, {4,3},{5,-1},{6,0}};

In[4]: BezierCurve[pts,t]Out[4]: {1+5t, 1+15t-70t2+140t3-140t4+54t5}

We remark here that we have obtained thesymbolic formula for the Bézier curve. This factrepresents a clear advantage with respect to thetraditional programming languages. Thesemathematical formulas are exact, because theycan be obtained by, more or less complex,symbolic operations. Since many computergraphics methods are very susceptible to round-off errors, this capability is specially valuable.

Plotting the corresponding curve reduces to thissimple input:

In[5]: PlotBezierCurve[pts,t]Out[5]: Figure 2(left).

Figure 2(left) shows a Bézier curve, its Bézierpoints and the control polygon (a set of segmentsjoining the control points of the curve).

Figure 2. (left) A Bézier curve; (right) Convex hullproperty of a Bézier curve

An interesting property of the Bézier curves isthe convex hull property: the curve lies entirelywithin the convex figure set by the extremepoints of the polygon determined by the controlpoints (see [1, 4, 6]). This fact is illustrated inFigure 2(right), where the grey area indicatessuch a convex figure. To get this figure, weintroduce the control points and obtain the plot ofthe corresponding Bézier curve:

In[6]: p1={{1/2,1},{2,3/2}, {3,1},{4,2},{5,1/2},{6,3}};In[7]: cur=PlotBezierCurve[p1,t];

Then, we load the Mathematica package forcalculating the convex hull of the given points.Finally, we display the resulting area (in grey)and the curve:

In[8]: <<DiscreteMath` ComputationalGeometry`

In[9]: ConvexHull[puntos];

In[10]: Show[ {Graphics[ {RGBColor[1,1,0], Polygon[%]}],cur} ]Out[10]: Figure 2(right)

In computer graphics, curves and surfaces areusually dealt with by using numerical routines,which are required to be as short, fast, elegant,numerically stable and general as possible. Forexample, the problem to evaluate a Bézier curveat a given parameter value t=t0 ∈ [0,1] (say t=0.3)is solved by applying the de Casteljau algorithm(see [3], Chapter 4). However, in our case thisproblem just reduces to compute:

In[11]: BezierCurve[pts,0.3]Out[11]: {2.5,1.97722}

3. We have used pattern recognition to identify thedimension of the points through the pattern {_,_}or {_,_,_}. Moreover, the pattern {{_,_}..}({{_,_,_}..} resp.) indicates that one or morepairs (triplets resp.) are expected. Otherwise, noevaluation is performed:

In[12]:r={{1,1},{2,4,3,2},{4,3}};

In[13]: BezierCurve[r,t] Out[13]: BezierCurve[{{1,1},{2,4,3,2},{4,3}},t]

Finally, the operator "|" (line 1 in Table 2) isused to indicate "this pattern" or "this anotherone". This means that the same commands areable to work with two- and three-dimensionalcurves and surfaces:

In[14]: PlotBezierCurve[{{1,4,4}, {2,2,3},{3,0,3},{3,4,0}, {4,5,2}},t] Out[14]:

Figure 3. Two different viewpoints of a three-

dimensional Bézier curve.

4. The graphical Mathematica capabilities can besuccessfully applied to visualization. They

include the rendering of parametric, implicit andexplicit curves and surfaces, colors, grids,different points of view (see Figure 3), shading,shadows, intersection algorithms, etc. In additionof this, the package incorporates many otheroptions, as two- and three-dimensionaltransformations, projections, perspectives, visiblelines and surfaces techniques, etc.

We remark that the previous examples have beenchosen primarily for simplicity and illustrativepurposes, rather than to correspond to valuable codesor very complicated algorithms. However, verysimilar discussions can be applied to many othercommands. As an illustration, Table 3 shows thecorresponding code for a Bézier surface. In this code,we have used the functional programming, apowerful alternative to the standard proceduralprogramming. Note that the pattern recognition andfunctional Mathematica programming avoidconditionals and loops, respectively. Motivated bythe different kinds of programming that Mathematicacan support, the final syntax of our commands isvery short, elegant and easy to understand.

BezierSurf[pts:{{{_,_,_}..}..},v1_,v2_]:= Module[{m=Length[pts]-1,U,V, n=Length[First[pts]]-1}, {U,V}=MapThread[Table[ Bernstein[i,#1,#2],{i,0,#1}]&, {{m,n},{var1,var2}} ]; Plus @@ (U.pts*V) //Simplify ]

Table 3. Code for the Bézier surface.

x y z x y z x y z x y z x y z x y z0 0 1 0 1 2 0 2 3 0 3 3 0 4 2 0 5 11 0 2 1 1 3 1 2 4 1 3 4 1 4 3 1 5 22 0 3 2 1 4 2 2 5 2 3 5 2 4 4 2 5 33 0 3 3 1 4 3 2 5 3 3 5 3 4 4 3 5 34 0 2 4 1 3 4 2 4 4 3 4 4 4 3 4 5 25 0 1 5 1 2 5 2 3 5 3 3 5 4 2 5 5 1

Table 4. Set of 3-D points of control.

The following picture shows a typical output for theBezierSurf command with the control points givenin Table 4 and stored in the variable pts3d:

In[15]: BezierSurf[pts3d,{u,v}]Out[15]: {4 v,2 u,2 v (3+(-3+u) v)}

The corresponding patch surface is shown in Figure4(left):

In[16]:PlotBezierSurf[ptos3d,{u,v}]Out[16]: Figure 4(left)

Figure 4. (left) Bézier patch; (right) B-spline patch.

Of course, many other commands (which are notdescribed here for limitations of space) for dealingwith the most usual curves and surfaces in CAGDhave also been implemented. Figure 5 shows twoMathematica palettes with the most importantcommands included in the package.

Figure 5. Some CAGD package commands.

In the next section these commands will be appliedto solve some interesting and illustrative examples.

3. Some Examples of Application

This section is devoted to show some examples ofapplication of the CAGD Mathematica packagedescribed in the previous paragraphs.

3.1. Two-dimensional transformationsIn this example we start with an initial square, givenby joining the list of points:

In[17]: square={{-5.,-5.},{5.,-5.}, {5.,5.},{-5.,5.},{-5.,-5.}};

Now, some two-dimensional transformations areapplied to the previous list (see [4], Chapters 2 and 3,for a survey on two- and three-transformations). Firstof all, we consider a proportional 2D-scaling (givenby the Scaling2D command) of factorSqrt[68]/10. Then, a rotation (Rotation2Dcommand) of angle arctan(5/3)- π/4 is applied. TheNestList command is finally applied to produce a40-steps iteration of this process, as follows:

In[18]:s=NestList[Rotation2D[ Scaling2D[#, {Sqrt[68]/10,Sqrt[68]/10}], ArcTan[5/3]-Pi/4]&,

square,40];

obtaining the Figure 6 by using the graphicalMathematica commands:

In[19]:Show[Graphics[ListPlot[#, PlotJoined->True, Axes->False,PlotRange->All] ]& /@ s,AspectRatio->1]Out[19]:

Figure 6. Iterative process of scaling and rotationtwo-dimensional transformations.

Figure 7. Mosaic obtained by the application of somesimple (rotations, scalings and reflections) two-

dimensional transformations.

Finally, reflecting and traslating Figure 6 (by usingthe Reflection2D and Traslation2Dcommands) Figure 7 is obtained. As the reader mayappreciate, the previous transformations can beachieved in a very easy and intuitive way, by thesimple application of the Mathematica and CAGDpackage facilities.

3.2. The Spinning Top ExampleGiven the list of points:

In[20]:spitop={{0,16},{1,16}, {1.5,16},{1.5,16},{1.5,16}, {1.5,16},{1.5,14.5},{1.5,14.5}, {1.5,14.5},{1.5,14.5},{1.5,14.5}, {2,14.5},{3.5,14},{5,13},{5.5,11}, {4.5,8.5},{3,7},{1.5,6},{1,4.5}, {0.5,3},{0,1.5},{0,0}};

the mathematical expression for its Bézier curve isgiven by:

In[21]:BezierCurve[spitop,t]

Out[21]: {21 t-105 t2+2992.5 t4-20349 t5+81396 t6-232560 t7+508725 t8- 881790 t9+ 1.2345 106 t10-1.23451 106 t11+146965 t12+ 2.34013 106 t13-4.593106 t14+ 4.17833 106 t15-1.30234 106 t16- 1.21196 106

t17+ 1.67114 106 t18-915600 t19+259717 t20-31662.5

t21, 16-81396 t6+1.04652 106 t7- 6.40993 106 t8 +2.46901 107 t9- 6.66633 107 t10+ 1.33327 108 t11-2.0384 108 t12+2.42764 108 t13-2.27734 108 t14 +1.69277 108 t15-9.98322 108 t16+4.6437 107 t17-1.666107 t18+4.36432 106 t19-737383 t20+59270.5 t21}

which is a polynomial of degree 21, obtained fromthe 22 control points (see Figure 8(left)). A 2D-reflection is then applied in order to obtain a wholespinning top, as shown in Figure 8(middle). Finally,several 2D-rotations can be applied for simulatingthe movement of the spinning top in a plane (seeFigure 8(right)). Finally, Figure 9 shows fourdifferent frames of a movie generated from the 3-Dtransformation commands implemented in thepackage.

Figure 8. Simulation of the spinning top movement:

three different steps.

Figure 9. 3-D movie of the spinning top.

3.3. B-spline CurvesIn general, Bézier curves are not appropriate forcomputer design because no local control can beperformed, that is, moving the position of onecontrol point alters the shape of the whole curve (seeFigure 10(left)). Moreover, the resulting curvedegree, n, is totally determined by the number of thecontrol points (see [1,3-6]). Hence, it is moreinteresting for computer graphics to consider B-spline curves, whose degree can be chosen and thatexhibit local control, that is, if a control point ismoved only the closest parts of the curve to thispoint are affected, as shown in Figure 10(right).

Figure 10. Different control for the curves: (left)Bézier: global control; (right) B-spline: local control.

As another example of B-spline curves, Figure 11shows three different pictures of a squirrel obtained

by using the PlotBSplineCurve command.They differ in the options, ControlPoints andControlLines, used to display a given picture,which can take the values True or False,indicating whether or not the picture is displayedwith the control points and/or lines.

Figure 11. Example of B-spline applied to design.

In addition of this, different choices of the knotsvectors, control points, and orders can be taken, forboth curves and surfaces. This will be illustrated inthe case of a surface in the next example.

3.4. B-spline SurfacesFigure 4(right) showed a typical example of a B-spline patch, which has been built as a tensor productof two B-spline curves (see [1, 3, 4, 5] for moredetails). In this case, a periodic knot vector has beenchosen for both axes. Compare with the Bézier patchdefined by the same control points (Figure 4(left)).

The implemented commands allow us to choosemany different options. As an illustration, Figure 13shows two B-spline surfaces: on the left, a (3,3)-order with two nonperiodic knot vectors surface; onthe right, a (5,2)-order with a nonperiodic and aperiodic knot vectors. Differences between them canbe visually stablished.

Figure 13. Two different choices of orders and knotvectors for a B-spline surface.

3.5. Blending SurfacesIf a single parametric surface is not able toapproximate a given set of points sufficiently well,then we may use several patches which are joinedtogether. Figure 14 illustrates this blending process.Two Bézier surface patches F0 and F2 (see Figure 14(left)) are connected using a Bézier F1 with C1

continuity. Following [3], pag. 269, this connectioncan be achieved by a careful choice of the controlpoints for the Bézier surface F1. Figure 14(right)shows the resulting blending surface.

Figure 14. C1-continuous connection of two Béziersurfaces.

Once again following [3], Figure 15 shows anotherexample of blending surface: in this case, it is givenby two different curves, a circle and a ellipse.

Figure 15. Blending an ellipse and a circle.

3.6. The CAGD Package in IndustryIn many industrial areas, geometric information isstored, transmitted and manipulated in differentstandard formats, like IGES, VDA, CATIA, etc.Each of these standard formats support a differentmathematical representation of the objects. Forexample, VDA uses the monomial representation,whereas IGES supports the B-spline one (see [7] formore details on the IGES format). Since the CAGDpackage is able to deal with this last description, wehave developed an IGES-Mathematica converter.This program allows us to have access to somegeometric entities of the electronic files from theautomotive or aerospace industries. After reading thefiles, we can apply all the above describedcommands to these geometric entities and introducethe students to the industrial world.

4. Interaction with New Computer GraphicsEducation Resources

Nowadays, the new technologies are changing theusual teachers' resources for computer graphics andvisualization education. Among them, we considerthe use of the virtual reality languages and Internet.

4.1. Conversion to VRML languagePerhaps one of the most recent innovations incomputer graphics education is the appearance of thevirtual reality languages. By using these languages,

like VRML, you can navigate around and throughthe graphic and manipulate it in a very easy andintuitive way. In virtual reality, the user "lives" in thescene. Evidently, this opens many possibilities, someof them still unexplored, not only technical but alsoin terms of students' motivation. E. Donley, at IUP,has developed a Mathematica package that convertsMathematica 3D graphics into VRML format andwrites it to a .wrl file. This file can be viewed using avariety of VRML browsers, like Netscape. Forfurther information, we refer the reader to [8]. Thisprogram allows the students a better manipulation ofthe graphics, adding many options not available inthe Mathematica outputs.

4.2. Use of InternetToday, many teachers are concerned about thepossibility to take advantage of Internet as a vehiclefor communication and education. Indeed, there is alarge collection of Java programs and Web pages forcomputer graphics education. In this sense,Mathematica 3.0 and higher also allow to convertany standard Mathematica notebook to HTML.Unfortunately, pictures are fixed as they areconverted to GIF and JPEG formats. However, thereis a free Java applet, LiveGraphics3D, by M.Kraus [9], for incorporating some kind ofinteractivity (display and rotation) to these picturesin your Web pages. Some examples can be found in[9].

5. Conclusions

In this paper we have introduced a Mathematicapackage for learning and teaching CAGD andComputer Graphics. This package constitutes, as faras we know, the first attempt to bring SCPs, andspecially Mathematica, to the Computer Graphicscommunity in an extensive way. This work arosefrom our feeling that SCPs could be successfullyapplied in computer graphics and visualizationeducation. After using the packages in some regularcourses, our experience was so satisfactory that weare convinced now that this is one of the best toolsfor introducing students to CAGD and computergraphics.

The weakest part to be appreciated in this work isrelated with the requirements in memory andcomputation time that Mathematica requires, incomparison with other programming languages. Onthe contrary, the Mathematica code is shorter andeasier to understand, and many mathematicalfunctions are already incorporated. For example, thestudents can immediately obtain a graphical

representation of the problem under analysis and itssolution without using any external visualizationprogram.

Acknowledgements

Authors would like to thank the CICYT (ComisiónInterministerial de Ciencia y Tecnología) of theSpanish Ministry of Education (project TAP98-0640), the European FEDER funds (project 1FD97-0409) and University of Cantabria for finnancialsupport of this work.

References

[1] V. B. Anand. Computer Graphics andGeometric Modeling for Engineers, JohnWiley and Sons, New York, 1993.

[2] S. Wolfram. The Mathematica Book. 3rd.ed., Wolfram Media / Cambridge UniversityPress, 1996.

[3] J. Hoscheck and D. Lasser. Fundamentals ofComputer Aided Geometric Design. A.K.Peters, Massachussetts, 1993.

[4] D.F. Rogers and J.A. Adams. MathematicalElements for Computer Graphics. SecondEd., Mc Graw-Hill, New York, 1989.

[5] L. Piegl and W. Tiller. The NURBS Book,Second Ed., Springer-Verlag, Heidelberg,1997.

[6] G.Farin. Curves and Surfaces for ComputerAided Geometric Design, Academic Press,Orlando Fl, 1988.

[7] IGES/PDES Organization. The InitialGraphics Exchange Specification (IGES).Version 5.1. National Computer GraphicsAssociation. Virginia, USA, 1991.

[8] E. Donley. VRMLConvert program. WebPage: http://www.ma.iup.edu/people/hedonley. html

[9] M. Kraus. LiveGraphics3D program. WebPage: http://theorie3.physik.uni-erlangen.de/~mkraus/Live.html

The Development of a Digital Library to Support the Teaching ofComputer Graphics and Visualization

G. Scott Owen, Raj Sunderraman and Yanqing ZhangDepartment of Computer Science

Georgia State UniversityAtlanta, Georgia USA

owen@siggraph.org

Abstract

The ACM SIGGRAPH Education Committee website is being evolved into a Digital Library thatwill be a worldwide resource for Graphics andVisualization Education. We discuss the history ofthe website, including a long-standing partnershipwith the National Science Foundation, and thecurrent work to evolve it into a Digital Library.Finally, we discuss a proposed vision of the futureof the Digital Library and how we plan to achievethat vision.Keywords: Digital Library, XML, graphics andVisualization education.

1. Introduction

The National Science Foundation (NSF) in theUnited States, in conjunction with other federalagencies, has started a major research initiative onthe design and creation of Digital Libraries. Aspart of this initiative, the NSF Division ofUndergraduate Education (DUE) has a vision of anational Digital Library to support education inScience, Mathematics, Engineering, andTechnology (SMET). The goal is to leveragepreviously funded NSF projects that havedeveloped curriculum materials, and to combinethe results of these projects, as well as future suchprojects, with research into Digital Libraries todevelop useful Digital Libraries to supporteducation in SMET. Since these Digital Librarieswill be based on the World Wide Web (WWW)they will be international in nature.

This paper addresses the effort to create andenhance a Digital Library to support education inComputer Graphics and Visualization. This is ajoint effort between the NSF and the ACMSIGGRAPH Education Committee. As part of theSIGGRAPH Education Committee web site(www.education.siggraph.org) we have had aComputer Graphics Courseware Repository(CGCR) for several years. This has now beenextended to Visualization, becoming the Computer

Graphics and Visualization CoursewareRepository (CGVCR). The CGVCR is beingenhanced and is evolving into the SIGGRAPHEducation Committee Digital Library (SECDL).This work builds on previous cooperative workbetween the NSF and the Education Committee.

There are two main tasks involved in the creationof a Digital Library to support education. The firstis the assembling and/or production of relevanteducational materials and placing them onto a website. The second task is the evolution of this website into a Digital Library.

2. Digital Libraries

A Digital Library differs from a database, website, or online repository in the way it is structuredand the way that users can find, access, andretrieve the information. As the amount ofmaterials increases, the ability to locate the neededmaterials, or even to browse the information,becomes more difficult. New and innovativemethods of navigation are needed. In addition, thematerials should be flexible enough to be reused orrepurposed, i.e., reconfigured and easily deliveredin different sized “chunks” of information. Finally,there should be a sense of community among thedevelopers and users of the materials. The visionof a Digital Library is that of a repository ofmaterials wherein all of this is possible.

Now let us apply this vision to graphics andvisualization educators. A professor teaching acourse in global illumination wants to findmaterial on ray tracing caustics. She goes to theSECDL, does a search, and retrieves teachingmaterials consisting of text, images, and videosillustrating the principles. She also finds relevantportions of SIGGRAPH Conference Course notesand links to original research papers that arehoused at the main ACM SIGGRAPH site(www.siggraph.org). She decides to visit the SECWorld (see later discussion) to see who might bearound. There, she meets an Art professor who, as

she discusses her work, decides that he is alsointerested in the same topic, but from an artperspective. He goes to the SECDL and retrieves aset of materials that also cover caustics, but thatare suitable for his art classes.

Another professor teaching visualization wants tofind images of all visualization techniques thatproduce images that have an appearance similar tostreamlines. So, he performs a query by imagecontent and finds a set of images, ranked insimilarity order.

Finally, a professor teaching human modelingtechniques wants to illustrate proper and realisticposes, but is unsure of just how to do it. She goesto the CGVCR and finds a section onContrapposto with images comparing Egyptianand Greek sculpture, as well as some byDonatello.

The common thread in all of the above is that theprofessor may have an exact idea of what hewants, or only a vague idea and may just bebrowsing. In all cases the SECDL is able tosupport their needs.

Computer Graphics and Visualization are uniquein that they are of interest to a wide variety ofdisciplines, e.g., engineers, mathematicians,computer scientists, natural and physical scientists,artists, graphics designers, animators, etc. Thus,educators from all of these disciplines need tohave access to materials to teach their own view ofthese topics. Ideally, the materials could be reusedand tailored to the different views.

3. Previous and Current Efforts onIncreasing the Materials in theCGVCR

Previous efforts were focused on building up theamount of educational materials available on theEducation Committee web site. The EducationCommittee and the NSF have collaborated onseveral previous projects to provide facultyenhancement workshops in Computer Graphics(1990 (NSF USE-8954402), 1993, and 1994 (NSFDUE-9255489), and Scientific Visualization (1996and 1997 (NSF DUE-9554692). While the primarythrust of these projects was the faculty workshops,some educational materials were created andplaced on the SECDL (HyperGraph andHyperVis). In addition, there has been somedonation of materials from other sources.

We have intensified our efforts to gather and/orproduce educational materials to place on theSECDL. A recent effort, funded by the NSF(DUE-9752398), has been a project to developinstructional materials to teach concepts ofVisualization. These materials are being developedin conjunction and accord with therecommendations of the Education CommitteeCommittee on Education for Visualization,Chaired by Gitta Domik, University of Paderborn,Germany. In June 1996, The U.K. Advisory Groupon Computer Graphics (AGOCG) lead by AnneMumford, held a workshop on computer graphicsand visualization education. Several of therecommendations of that workshop are beingimplemented in this project. In addition, KenBrodlie, University of Leeds, was funded byAGOCG to develop a web site on VisualizationTechniques, which has now been added to theSECDL.

We have begun to place more materials from theannual SIGGRAPH Conference on the SECDL,e.g., Educational Slide sets, and some SIGGRAPHConference Course notes. We plan to put papersfrom the Educators program on the site. We alsoplan to place some digitized versions ofSIGGRAPH Video Reviews on the SECDL. Ofcourse, all of the above depends upon gettingproper copyright permission from the originalauthors. We also have been in contact with theorganizers of the annual IEEE Visualizationconference and will try to place some of thesetutorials, etc. on the SECDL.

In summary, the SECDL has a rich set ofmultimedia learning materials. However, the veryrichness and size of the SECDL, and the differentviewpoints and needs of the educators that want toaccess it, makes the material not as accessible andflexible as desired, which is the motivation forthese projects.

4. Digital Library Project

We have just begun another set of projects, incollaboration with the NSF to evolve the CGVCRinto the SIGGRAPH Education Committee DigitalLibrary (SECDL). As discussed above, thisprimarily means improving the navigation andretrieval capabilities and the flexibility of thematerials. Some of this work is currently fundedby the NSF DUE-9816443), while some has beenproposed and we have not yet begun work on it.

4.1. Current SECDL ProjectsThese projects involve improving the navigationand information retrieval aspects of the SECDL.We are approaching this in several ways. We haveimplemented a conventional search engine(InfoSeek). However, this type of technology islimited so we are performing research on twoother types of navigation and retrieval.

It is the opinion of many people that databasetechnology will play a critical role in DigitalLibraries. However, there is one importantdifference between the information available onthe Web and the information usually stored in atraditional database: the lack of "structure" in theWeb information and the strong "structure" to theinformation stored in a database. Recently, theterm "semi-structured" data has been coined torefer to the data and information such as thosestored on the Web. The SECDL falls under thiscategory of "semi-structured" information. We areinvestigating two different data base systems thatdeal with semi-structured data on the Web. Weplan to use these systems in conjunction with theSECDL to provide ad-hoc querying and searchcapabilities.

One system is Lore (Lightweight ObjectRepository, a new database system, developed atStanford University, designed to support storageand queries of semistructured data. In Lore, data isself-describing, so it does not need to adhere to aschema fixed in advance. Lore is particularly wellsuited for document data, including HTMLdocuments available on the World Wide Web.

The second approach is Virtual databasetechnology, a radically different approach tomanage and query semi-structured data on theWeb. In this approach, textual data is converted tostructured relational database data and queries areperformed on the relational database. All this istransparent to the user who is interacting with aWeb agent which receives search and queryrequests and processes them by invoking dataconverter modules (adaptors) and then querying arelational database.

Virtual database technology transforms the Webinto a database, using adapters that analyze theHTML from Web sites to present them asrelational data sources. The limited descriptivecapability of HTML necessitates manual analysisfor the creation of adapters. However, with XML(see later discussion), the new standard for Web

publishing, the creation of these adaptors can beautomated.

We are also investigating Visual InformationRetrieval techniques, such as the Query by ImageContent (QBIC) system from IBM and the Oraclebased VIR system. These systems will allow usersto use graphical queries, e.g., either start with anexisting image or sketch a new image and requestsimilar images. A user will also be able to searchfor videos that have parts that match a certain typeof motion, as drawn by the requester.

4.2. Proposed SECDL ProjectsThese projects have been submitted as proposals tothe NSF Digital Libraries Initiative and, if funded,will commence in January, 2000. We have done,and are continuing, some preliminary work on theprojects. These projects extend the above efforts toincrease the navigation and information retrievalaspects of the SECDL and also directly address thereusability problem. They also include the creationof SEC World. These projects involve XML itsassociated technologies.

The eXtensible Markup Language (XML) [1] is amarkup language for documents containingstructured information, i.e., information thatcontains both content (words, pictures, etc.) andsome indication of what role that content plays. Itis different than HTML, which is a markuplanguage for the display of the documentinformation but which specifies nothing about theinformation content. The XML specificationdefines a standard way to add markup todocuments. Unlike HTML, XML is extensible,i.e., you can create your own tags.

XML can be applied to any application area bycreating a Document Type Definition (DTD) forthat topic. The DTD is the set of XML tags for theparticular area and is specific to that area. DTDsare the technology that allow standard XML tagsto be defined, and that permit different platformsand machines to understand XML descriptions andapplications. DTDs for specific disciplines havebeen previously created, e.g., for mathematics [2],chemistry [3], and biology [4].

4.2.1. Material ReuseFor this project we will create a DTD for computergraphics and visualization. This DTD will bebased on the results of the SIGGRAPH EducationCommittee Graphics Taxonomy Project, led by

Jacquelyn Ford Morie. The goals of the TaxonomyProject are to create a complete Taxonomy forComputer Graphics and Visualization and to haveeducators from different disciplines createdifferent curricular views of the topics. This DTDwill allow us to semantically markup anydocument in these areas.

Another important part of XML technology is theeXtensible Style Language, or XSL. An XSL stylesheet is an XML document containing a set oftemplate rules. These template rules describe howthe original XML element node is converted intoan XSL element node that can be formatted,styled, and displayed. By having different XSLstyle sheets, a set of XML documents can betailored to different views. We will create differentXSL style sheets in our project that correspond tothe different curricular views of educators wantingto use the resources in the SECDL.

There is a huge effort to produce XML tools inindustry and we plan to leverage this effort. Thesetools are all new and in a state of rapid flux. Overthe next six months many more such tools willbecome available. An example is the IBM site(http://ibm.alphaworks.com) which has a set ofXML tools that are freely available.

Even though the Taxonomy Project will not becompleted until August 2000, we can begin todevelop and test prototype DTD’s based on thecurrent version of the Taxonomy Project at theanticipated start of this project (January 2000). Asa feasibility study, we have already begun work onthis using the current version of the Taxonomy.We have taken part of the section on Rendering(Lighting Models and Texture Mapping) andcreated a prototype DTD from this. We have foundthat this is a straightforward process, enough togive us the confidence that, given the final, fullTaxonomy, we will be able to create a full DTD.

4.2.2. 3D Navigation of the Information SpaceNavigation in a complex 3D space is a difficultproblem. The user needs to be able to have anoverview of the complete space and be able to divedown into the details of a particular subspace. Theuser should also have a trail and know which partsof the space they have already visited. This is nowpossible with the 3D Graphics technologiesassociated with XML. There has been someinterest in relating XML and VRML technology.There has been a draft DTD created for VRMLcreated by Bullard [5]. Lipkin [6] discusses the use

of XSL Stylesheets to dynamically generateVRML worlds to display the results of databasequeries. Ancona [7] discusses 3DXML as a way ofdescribing websites and other structuredinformation spaces in XML and publishing themin VRML. This would allow for 3D navigation ofthe information space. For the relationshipbetween XML and 3D Graphics, see also [8].

The objective of this portion of our project is to beable to generate a 3D VRML world that willrepresent the particular view of the informationspace generated in the first part of the project.Assume a user in computer science has requested aparticular view of a topic area. From the first partof the project, this query will have generated a setof XML/HTML pages and associated materials.This query will also generate a 3D VRMLrepresentation of all of the materials in the form ofa hierarchical graph structure.

In this VRML world each node might be a spherethat could contain a sub-graph structure. Sub topicareas could be represented as translucent spheresthat the user could move into as they traveled.There would be a nested graph structure that couldrepresent an information space of considerablecomplexity. For example, a top-level node mightbe Rendering. This might then contain a sub-graphwith one of the nodes being Lighting Models. Thiscould contain another sub-graph with nodesrepresenting Local Illumination and GlobalIllumination models, each with additional sub-graphs. The user can navigate in this VRML worldto find the different types of materials. The usercan pull back and look at the entire space ortraverse along the links to look at differentvolumes of the space.

When the user visits a portion of the space thatnode/sphere would be marked, e.g., a colorchange, as having been visited. Since a user canhave more than one instance of a VRML browser,they could simultaneously have multiple views ofthe information space world.

4.2.3. Sense of CommunityDeveloping a sense of community among DigitalLibrary users and developers was one of theimportant issues identified at the NSF January,1999 workshop on Digital Library research issues.There are several ways we have addressed this inthe SECDL. The first is a simple listserve that isused for announcements and has some discussion.But a listserve or web discussion page is

asynchronous and not real time. A real time chatroom engenders more sense of community, but istext based and not visual. We propose to develop a3D virtual world that our community will be moreinterested in using.

Imagine the existence of a fully 3D interactivevirtual world (SEC World) and the followingpossible scenarios:

A computer science professor from California logsin to SEC World. As she wanders around sheencounters another avatar who is an art professorfrom Ohio and also a professor from theUniversity of Leeds, in the U.K., who teachesvisualization. They engage in a 3-way discussionabout the possibilities of incorporating aspects ofeach others’ disciplines into their own courses.The art professor mentions a book on drawing thatillustrates how certain techniques can help aperson focus in on what is important in an imageand the Leeds professor decides he will look at itand incorporate parts of it into his class. Afterfurther discussion the three of them decide to gettogether later and submit proposals to theirrespective Universities for interdisciplinarycourses that they will jointly develop.

An engineering professor from Colorado entersSEC World for the first time. He is just beginningto teach courses on Engineering Graphics and he isnot sure what he is looking for and would like tofind out what all this is about. He sees an avatarwith a baseball cap that identifies her as an SECWorld Guide and goes and asks her somequestions. (He knows that she is a guide becausethe Student Volunteers at the annual SIGGRAPHConference wear baseball caps to identify them.)She also gives him brief descriptions of the partsof SEC World and answers some of his questions.He thanks her and goes to another room whereanother engineer is talking with some artists.

The people in the above cases might not be facultybut could just as easily be students who arelooking for some interdisciplinary help that mightnot be readily available at their own institution.

The above is possible by the use of commerciallyavailable Multiple User Shared Environment(MUSE) worlds. SEC World could be created anddownloaded the first time from the centralSECDL. After that, the users could invoke thisworld and enter it without downloading a largeamount of information. A complex world needs apowerful processor and graphics card, but these

are becoming very inexpensive. The requiredbandwidth is not very high since the onlyinformation that is being transmitted is thechanging coordinates of the avatars and the textchat. If the users use microphones, then thebandwidth requirements are higher, but still notexcessive. These commercial worlds are designedfor users who communicate to the Internet via28/56K modem.

Users could choose an avatar in the database oreven design a new one and upload it to the server.The special guide mentioned above would not be aperson but an intelligent agent who would haveaccess to a large amount of information about theSEC World as well as the SECDL. We are notproposing a full natural language discourse butonly a subset.

We have experimented with both SonyCommunity Place and the Blaxxun [9] communityand found that they are similar. If we create aVRML world for one then it is trivial to convert itto the other format.

4.2.4. EduCause IMSA final proposed project relates again tonavigation and retrieval issues. We have been incontact with the developers of the EduCauseInstructional Management System (IMS), an XMLbased system of metadata for general instructionalnavigation and retrieval. EduCause is proposingthat IMS become a standard for all educationalsites. We are planning on using the SECDL to actas a multimedia testbed for IMS version 1.0. Wewill extend the general IMS metadata to be morespecific for topics in graphics and visualization.This project is a collateral project with thosediscussed above and has also been submitted toNSF as part of a consortium effort led by theCollegis Research Institute.

5. Conclusion

The evolution of the current CGVCR into a fullDigital Library, the SECDL, will provide muchgreater functionality for graphics and visualizationeducators. Using new XML and relatedtechnologies, we plan on increasing the flexibilityand navigation capabilities of the SECDL andenhancing the sense of community among itsusers. . The SECDL has the potential to be aninternational resource that can service educatorsworldwide. There are many research problems thatmust be investigated and overcome before the

ideal can be realized, however once achieved, theSECDL will provide a prototype of the educationalDigital Library of tomorrow.

References

[1] See for example Natanya Pitts-Moulis andCheryl Kirk, XML Black Book, CoriolisTechnology Press, 1999; See alsohttp://www.xml.com;http://www.xmlinfo.com

[2] http://www.w3.org/Math[3] http://www.venus.co.uk/omf/cml/ also see

Peter Murray-Rust “Chemical MarkupLanguage: A Simple introduction toStructured Documents”http://www.xml.com/xml/pub/w3j/s3.rustintro.html

[4] http://www.visualgenomics.com/[5] Len Bullard, “Draft VRML DTD”,

http://ariadne.iz.net/~jeffs/vrmLab/Documentation/vrmlDTD.html

[6] Daniel Lipkin “Integrating XML andVRML: A Technical Discussion”http://www.vrml.org/WorkingGroups/dbwork/vrmlxml.html July 30, 1998

[7] Dan Ancona, “3DXML Authoring,Publishing and Viewing StructuredInformation”,http://www.vrml.org/WorkingGroups/dbwork/ancona/home.html, August 17, 1998

[8] Web3D Consortiumhttp://www.vrml.org/fs_specifications.htm

[9] Blaxxun Corporationhttp://www.blaxxun.com

Learning 3D Visualis a tion with MALUDA

João A. Madeiras Pereira and Mário Rui GomesInstituto Superior Técnico/Instituto Engenharia Sistemas e Computadores

Lisboa, Portugal.{jap, Mrg}@inesc.pt

Abstract

This paper describes MALUDA system, whichconstitutes an important tool for teaching 3D-visualisation pipeline at the laboratory classes of theComputer Graphics course at Instituto SuperiorTécnico.Keywords: application tool, 3D visualisation pipeline,modelling, rasterization algorithms.

1. Objectives

The traditional 3D-visualisation pipeline is one of themain topics to be taught in any introductory ComputerGraphics course. The MALUDA system constitutesitself as an important complementary tool to learnabout 3D visualisation. In fact, it is used at thelaboratory classes of the Computer Graphics course atInstituto Superior Técnico, not only as a way topractice the use of a virtual camera model based-viewing pipeline but also as a test environment forrasterization algorithms developed by the students.

2. MALUDA main features

There are three main approaches a novice student inComputer Graphics can take to make effective use ofMALUDA:

• Learning to create 3-dimensional scenes• Understanding and observing different realism

levels of image rendering• Testing his own scan conversion strategy.•

2.1. Three-dimensional modellingA scene can be built up by using the NFF (Neutral FileFormat) language and/or the SIPP (Simple PolygonProcessor) scene description library.

NFF [1] is a low level scene description language inthe sense that it provides a minimal interface whichallows a programmer to quickly create a scene bydescribing the geometry and basic surfacecharacteristics of objects, the placement of lights, andthe viewing frustum for the eye. There is no

hierarchical modelling. An example of a NFF file isillustrated in Appendix A.1.In addition to support the NFF scene descriptionlanguage, MALUDA also supports SIPP scenedescription library [2]. A SIPP scene is built up ofobjects (described in their local coordinates systems)which can be transformed with rotation, translation andscaling. The objects form hierarchies where each objectcan have arbitrarily many sub-objects. It allows also ascene to be illuminated by an arbitrary number of lightsources of different types. In a SIPP scene is possibleto create several virtual cameras, and then specify oneof them to use when rendering the scene.

An important feature of MALUDA modellingcapability is the possibility of merging a SIPP scenewith one described in NFF language. Then, a user canask either to render the resulting scene or to export it inNFF format (containing optionally only triangles).

2.2. Image RenderingMALUDA provides the ability for creating 3-dimensional scenes and rendering them using either theinternal scan-line z-buffer algorithm or an externalrasterization algorithm provided by the user. Bydefault, the internal rasterization engine is used in therendering operation. This engine, through the use of thepublic domain SIPP library, renders images at differentrealism levels. Scene objects are rendered with eitherPhong, Gouraud or flat shading. An image can also berendered as a line drawing of the polygon edgeswithout any shading at all. The SIPP library alsoprovides 3-dimensional texture mapping withautomatic interpolation of texture coordinates. Simpleanti-aliasing can be performed through oversampling.Transparencies are allowed and several shadingfunctions (“shaders”) are available. Some examplesgenerated by MALUDA by using the internal SIPPrasterization engine are illustrated in Appendix B.

2.3. Rasterization supportA major feature of this tool is the ability for a studentto provide his own rasterization function. This makes iteasy to experiment with various scan conversionschemes and to do special effects.

Two procedures can be used accordingly to theplatform to execute the rasterization algorithm isrelevant or not for the test. If execution platform is notimportant, the routine containing the rasterization codedeveloped by the student can be integrated intoMALUDA system to be invoked later. This procedureis meant to check the strategy and its performancewhatever the machine and operating system. Or, theplatform is important, like a parallel machine where weare interested to experiment several parallel schemesfor the rasterization stage. In this case, MALUDA

should be instructed to output an ASCII file containingdevice dependent coordinates polygonal database,which will be then used to feed the target rasterizationalgorithm. An example of this “flat” file is shown inAppendix A.2

3. MALUDA architecture

In figure 1, is depicted the MALUDA activity flowdiagram

Figure 1 - MALUDA activity flow diagram

NFF FileScene

Descript ion

World, View &Light definitions

(SIPP 3D Library)

Load DatabaseHierarchical Object

Structure

Object ModelingSurfaces, Polygons &

Geometr ic Transformat ion

Export DatabaseNeutral Fi le Format (NFF)

Export ?

Cul l ing ? Backface Removal

TransformationLocal to Wor ld

Coordinate System

Calculate PolygonNormals

Calculate VertexAveraged Normals

Polygon Clipping(Suther land-Hodgeman)

Transformation Perspective

Phong Reflection Model(Vertex Level)

Split ?Polygon to

Triangle Split

TransformationDevice Dependent

Coordinates

Export Rendering Data(Coordinates & RGB) Export ?

Render ?Rasterization

Hidden Surface Removal ,Interpolat ion & Shading

ImageFile

RenderingFile

NFF Parser

Split ?

Polygon toTriangle Split

UserRasterization

Algor i thm

Yes

Yes

User

9

11

4

1

2

3

7

5

6

8

10

13

12

14

15

17

16

Yes

Yes

Yes

geométricasTransformações

Modelação e construção da base de dados(importação e exportação)

Gerador de saída

Internal

An analysis of it shows to a reader the most importantmodules and their associated functionality. Thus,accordingly to the MALUDA system description madeabove, we can distinguish the following functionalblocks:

• Reader and parser of scenes created by NFF and/orSIPP languages

• Geometric transformer• Output generator

At this point, is convenient to remind some importantfeatures of MALUDA and relate them with thenumbered operations illustrated in Figure 1.

MALUDA can export the input scene as a NFF fileformat (World Coordinates) –operation 6– or as anASCII polygonal “flat” file format (Device DependentCoordinates) –operation 15. The system provides alsothe ability of image rendering by using either theinternal SIPP rasterization engine –operation 17- or anexternal rasterization algorithm provided by the student–operation 16.

It is interesting to mention that when MALUDA isconfigured to act as image renderer, diagnostic andstatistics informations are provided by the system.Appendix A.3 shows the information provided by thesystem when the default SIPP rasterization engine isbeing used.

References

[1] Eric Haines, “A Proposal for StandardGraphic Environments”, in IEEE ComputerGraphics and Applications, Vol 17, nº11,November 1987, pp 3-5

[2] J. Yngvesson, I. Wallin; “User’s Guide toSIPP –a 3D rendering library”, version 3.1,Public Domain Software via anonymous ftpfrom isy.liu.se (IP nº 130.236.1.3) in thedirectory /pub/sipp; March 1993

Appendix A – MALUDA Input/Outputinformation

A.1 NFF File Formatv Viewpoint definition;from 1.02285 -3.17715 -2.1745at -0.004103 -0.004103 0.216539up -0.816497 -0.816497 0.816497angle 45hither 1resolution 1024 1024l 1.87607 -18.1239 -5.00042 light source position;f 1 0.2 0.2 1 0 100000 0 0 surface parameters;p 3 Polygon with three vertices;-1 -1 1-1 -0.9375 0.9375-0.9375 -1 0.9375p 3-0.9375 -0.9375 1-0.9375 -1 0.9375-1 -0.9375 0.9375p 412 12 -0.5-12 12 -0.5-12 -12 -0.512 -12 -0.5f 1 0.9 0.7 0.5 0.5 45.2776 0 0 New surface;s 0 0 0 0. 5 a sphere;s 0.272166 0.272166 0.544331 0.166667…

A.2 Device Dependent Coordinates Output File512 512 65535 255 Maximum resolution (x,y,z and colour);120838 Total number of polygons;1 Separator;3 100 50 Polygon with 3 vertices and Ymax/Ymin;40 100 89 30 128 30 (x,y,z) and (RGB)50 80 89 30 128 3060 50 89 30 128 3014 110 40 Polygon with 4 vertices and Ymax/Ymin;40 40 89 30 128 3050 110 89 30 128 3060 50 89 30 128 3055 60 89 30 128 301127 End of file;

A.3 Diagnostic and Statistics informationNFF scene will be imported!Pipeline: Parsing NFF file, wait...Pipeline: Be aware that some SIPP and user defined view parameters will be override by NFF parameters!Pipeline: NFF background loadedPipeline: NFF view parameters loadedPipeline: NFF light loaded (6.00 6.00 6.00) (1.00 1.00 1.00)Pipeline: NFF surface with 1 polygons(0.30,0.60,0.30,5,1.00,0.85,0.70)Pipeline: NFF surface with 1 polygons(0.30,0.00,0.20,5,1.00,0.61,0.41)Pipeline: NFF surface with 145 polygons(0.30,0.00,0.20,5,1.00,0.61,0.41)Pipeline: NFF surface with 1 polygons(0.30,0.00,1.00,5,1.00,1.00,1.00)Pipeline: NFF surface with 145 polygons(0.30,0.00,1.00,5,1.00,1.00,1.00)Pipeline: NFF surface with 1 polygons(0.30,0.00,1.00,5,0.02,0.67,1.00)Pipeline: NFF surface with 145 polygons(0.30,0.00,1.00,5,0.02,0.67,1.00)Pipeline: NFF surface with 1 polygons(0.30,0.00,1.00,5,0.21,1.00,0.01)…Pipeline: NFF surface with 145 polygons(0.30,0.00,1.00,5,1.00,0.26,0.75)

Pipeline: Imported 129 surfaces with 9345 polygons, 0 spheres and 0cones.Pipeline: NFF parser donePipeline: Hither:1.00,Yon:15.00,Resolution(X,Y,Z,Color):1024,1024,1024,512Pipeline: Pipeline creating render data, wait...Timer started Geometric Transformations start;Pipeline: Image file to be created: 'pipe.ppm'Pipeline: Objects loaded: 3Pipeline: Surfaces loaded: 129Pipeline: Polygons loaded: 9345Pipeline: Polygons ready to be rendered by SIPP! GeometricTransformations end;Timer Stopped (Elapsed: 2.108s, User: 2.023s, System: 0.066s)Pipeline: SIPP rendering, wait...Timer started Rasterization stage;Pipeline: Render Mode: Gouraud Método de sombreamento;Pipeline: SIPP rasterization done!Timer Stopped (Elapsed: 112.693s, User: 108.058s, System: 1.053s)

Pipeline: Successful EndPipeline: Polygons backfacing: 4864Pipeline: Polygons totally clipped: 39Pipeline: Polygons splited to triangles: 0Pipeline: Polygons rendered: 4442Pipeline: Releasing resources...!Pipeline: Release complete

Appendix B – Different types of image rendering with SIPP rasterization

Teapot - Wire-Frame model Teapot – Flat shading

Appendix B – Different types of image rendering with SIPP rasterization (cont.)

Teapot –Gouraud shading Teapot – Phong Shading

Mount with transparent spheres Teapot clipped by Sutherland-Hodgman

Teaching the Graphics Processing Pipeline:Cosmetic and Geometric Attribute Implications

Jack Bresenham, Ph.D.Winthrop University

Rock Hill, SC 29733 USAbresenhamj@winthrop.edu

Abstract

The pipeline processing model for computergraphics typically includes stages for interpretingthe display list <object entity descriptions, picture

abstraction>, geometric manipulation <scaling,

rotational, skewing transformations>, raster rendering<scanline conversion, dot image approximation>, andfinally presentation <scanline refresh, image display>.At what point in the graphics processing pipelinedo attributes actually become associated withpicture entities? The answer depends upon thetype of attribute and the specification intent for theattribute. I believe it is instructive for students toovertly and consciously become aware attributessuch as line width and line style can have variousinterpretations. Cosmetic attributes are appliedpost-transformation; geometric attributes areapplied pre-transformation. Two different types ofclipping also can be considered as scissor clippingin post-rasterization discrete pixel space or asanalytic clipping in pre-rasterization continuousspace.Keywords: computer graphics attributes:cosmetic & geometric

1. Introduction

Is line width a purely cosmetic effect meant tomimic a finely sharpened ‘thin’ pencil or widelyworn ‘fat’ pencil line thickness? In this case, linewidth is associated with geometric entities post-transformation. Engineering line weights arecosmetic attributes applied by a draftsman.Cosmetic attributes do not scale, skew, orotherwise distort. Asking for an architecturaldrawing to be redone on a paper twice as long andtwice as wide treats line width as a cosmeticattribute since pens or pencils used will be thesame as those used in the smaller originalrendering. Pen nib selection or choice of pencillead width determines line thickness in thiscosmetic attribute instance.

Is line width a geometric property inherent in theoriginal display list abstraction meant to mimic

photographic enlargement? In this case, line widthwill be associated with geometric entities pre-transformation. Having created areas for formerlyzero-width lines, those areas must be processedthrough the full transformation stage with theeffect that line width area can be scaled, skewed,or otherwise distorted. Pressing the enlargementby 2 button on a xerographic copier applies linewidth as a geometric attribute. Use of an overheadprojector to produce large lines on a viewingscreen some distance away from the projector isanother example of a geometric interpretation ofthe attribute line width. Drawing or writing on aninflated balloon then letting much of the air escapeis a 3-D example of a shrinking figure size,geometric width reduction.

Do dashes in a line style transform from rectanglesto parallelograms with sharp points? Yes, they can,if one elects to apply line style as a geometric orpre-transformation attribute. Treated as a cosmeticor post-transformation attribute dashes will keeptheir rectangular shape. The basic question toresolve in one’s reference model is whether apicture is first conceptually rendered, then hasattributes applied, and is subsequently transformedand displayed or whether a picture is firsttransformed, then has attributes applied, and issubsequently displayed.

Both approaches are valid. Students shouldappreciate the subtleties and interactions involvedin each reference model approach for attributeapplication. I prefer the more mnemonicnomenclature I first heard used by Dr. BrianMiddleton: ‘cosmetic’ and ‘geometric’ attributes.

2. Motivation

Presenting simple line style implementationalternatives offers a good opportunity to reinforcetransformation concepts. I use rounded,semicircular end caps that transform into ellipticalcaps for dashed lines as a simple example. One canreadily derive the equation for the resultingellipses …but it can be messy to implement in

graphics hardware or software. If oneapproximates an original semicircle by chordsalong its perimeter, then, since lines alwaystransform into lines, implementation is facilitatedat reduced cost with concomitant reduced accuracy<fidelity>. The chords do introduce an earlyapproximation; all approximation errors thensubsequently propagate with the possibility thatrecognizable, visual degradation can be the result.

Clipping is another processing stage that can havealternative reference models. Should one firstraster, at least conceptually raster, the full picturein infinite raster space then cut out the smallviewing window to be actually displayed? Or,should one first clip the picture abstraction entitiesto an appropriately small viewing window andthen raster that post-clip result. The latter iscommonplace but it, too, introduces approximationerror earlier than need be the case.

A fairly easy demonstration of the two clippingeffects can be to first raster a full screen display ofa picture. Blank out a small rectangle, perhaps fora pop up menu, then clip the abstract picture toretain just the portion for this small blankrectangle. Finally, render that newly clipped smallpart of the abstract picture to heal and restore theoriginal full display picture. Discontinuities, likely,and undesirable mismatches off by a few pixelswill be visible in analytic clipping.

Alternatively, consider blanking the same area forthe small rectangular menu. This time, though,let’s assume a fully rastered copy of the originalpicture has been retained in a second,nondisplayed buffer. Bit-Blit <copy> the pixelstemporarily subsumed by the menu back into thedisplayed picture to restore the area overwritten bythe small menu. Likely the result is different fromthe previously described restoration process; thepicture will seamlessly be restored with nodiscontinuities in scissor clipping.

Here I like the term used by Dr. Bryan Roberts:scissor clipping to describe the model in whichone first conceptually renders a pixel space picturethen clips as is the case with the Bit-Blit approach.The alternative reference model in which one firstclips the picture abstraction in continuouscoordinates before rendering, then paints theclipped entity in discrete pixel space I refer to asanalytic clipping.

The graphics pipeline can be used to demonstratedifferent visible results one can obtain even for the

simple attributes line width, line style, andclipping. Students can try a few examples to seefor themselves how important a goodunderstanding of one’s underlying reference modelis. The alternative models also offer opportunity todiscuss introduction of early approximations withattendant error accumulation build-up.

3. Attribute Reference Model

Be fore imple me nting any gra phics inte rface , it is es se ntial to have a clea r c onceptual model in mind.Othe rwise , w ell-mea ning individuals may inte rpret acommon specification differently owing toprec onc eived notions of whe re in the graphic sproc ess ing pipeline an individua l ass umes attributeas socia tions operations take pla ce .

Clipping produces different visual re sults depe ndingupon whether one pe rforms a mathematica l c lip oftheoretic al, z ero w idth loc i of ge ome tric entityboundarie s or whe ther one first ra ste rs then sc is sorclips the re alize d pic ture in ra ster space . A differentpicture w ill be dra wn by thos e w ho cons ide r a logicoperation such as e xclus ive or to be applied afte r a polyline is drawn c onc eptually in an off-s creen s pac ethen exclusive or’d to the displayed picture framebuffer. Thos e who c ons ider logic operations to applyin a n on-scree n pixel by pixe l manner depe ndentupon the algorithmic s equence by w hic h the graphic s sys te m genera te s geometric e ntities duringre ndering will typically ac hieve a differe nt visualeffe ct.

A prope r referenc e model should trivially le t usansw er questions such as : D o circular dots a ndre ctangular da she s in a line style transform fromcirc les to e llips es and from rec ta ngles topa ra lle logra ms ? D o wide lines join smoothly or willmite ring be appropriate eve n though s ha rp spike sma y res ult a t the joining point? W ith a goodre fe rence mode l a nd agre ed upon stage s at whichattribute ac tions a re to be implemented in thegraphic s proce ssing pipe line, if w e don't like theansw er, w e s hould be a ble to cha nge the re fe rence mode l to achie ve the res ult w e rea lly inte nded.

4. Example: Line Width & Line Style

Le t's e xa mine a few implica tions of two poss iblere fe rence mode ls that ca n be use d in pipelineproc ess ing of Line Width and Line Style andinte rac tion with ge neral transformation and dra wingmode . I'll c all my alternatives: 1. G eometric <S tr etched Rubber S heet>

2. C os metic <Mul tiple Pen Nib S izes>

Line s a re spec ified by two end points s uch a sLINE(x0,y0,x1,y1) for the line from (x0,y0) to (x1,y1).Line s a nd any other ge ometric entity ca n betransformed by rota tion, sc aling, tra ns lation a ndshea ring by us e of a c oncatenate d general 2-Dtransformation of the form:

The res ult for any affine tra nsformation of a line isanother line s o, abstrac tly a t lea st, w e w ould ha ve aline from (u0,v0) to (u1,v1) as the res ult oftransformation. W ha t w e don't explicitly know is the effe ct expec te d from tra nsformation of eithe r linewidth or line style . The visual effec t achie ved byapplying attributes be fore trans forma tion ca nproduce distinctly diffe rent pic tures from that w hic hwould res ult from a pplying attributes only a fte r alltransformation is c omple ted.

Line width typica lly is taken to be perpendicular tothe defining line . Often spec ifica tions for line widthfa il to s pec ify e nd point s ha pe for a re a c losure.What, the n, shall w e tra nsform w he n dea ling with aline width s ince, in rea lity, we must trea t are as ?What othe r e ffects interact w ith the proce ss ; w ha tshapes are e mployed in line s tyle?

Graphic primitive s <l ines, r ect angles, el li pti cal arcs, ...>

ente r the pipe line with defining points in a n a bs tra ctworld c oordina te space . The defining points imply acomplete geome tric loc us whic h repres ents theinfinitely thin, ge ome tric abstrac tion of the gra phicprimitive form <contour or outli ne> in original, pre -transformation spac e. The e ss enc e of the pipeline proc ess ing is to colle ct as socia te d a ttributes <suchdecl arati ons as l ine width, l ine styl e, dr awing m ode, and clippi ngvalues> the n to rende r c onceptually the geometric form. The firs t c onceptual re nde ring of the full loc us in a bstra ct model s pac e is then subje cted to an a ffine transformation <whi ch can be the cumulative resul t of

concatenation of multi pl e t ransf or mat ions> for composite sc aling, rotation, tra ns lation, and s he aring. The re sulting post-transformation, c onceptually abs tract,picture obta ined from the loc us trans forma tionpote ntially is clipped before then be ing mapped tode vice coordinate s pac e.

When line width is implied in a ge ome tric se nse ,graphic primitive s suc h as a line, re ctangle or e llips einhe rently have a n are a definition. In the a bse nc e of

clos ure of the area ow ing to lac k of line end s ha pespec ifica tion, we c ould consider various pos sibilities for width proc ess ing. An impleme ntation ca n definean e nd shape or e nd ca p suc h as axially shea red,pe rpendic ula rly butted, semic irc ular, or triangular.For a perpendicular butt, w e may w ant to use either atrue end point re fe rence or a n e xtended end pointre fe rence in a nticipation of joining line se gme nts toform, s ay, a right angle corner.

With an a nis otropic transformation, s emicirc ula r endca ps ca n be se en ultimately on the display s urfac e a spa rtial e llips es ca pping ends of line s. As a n e xa mple,try a refere nc e line s egment from (0,0) to (4,0) witha line width of two units on eac h side of there fe rence line . Let transformation ma trix pa rameters be : a=2, b=2, c=3, d=1, e=0, and f=0. A pre-transformation se micircle c entered at the origincould be the leftmost end c ap joining the 'w ide line 'end points (0,2) and (0,-2); a c omparable se mic ircle ce ntere d at (4,0) c ould join the rightmost e nd ca ppoints from (4,-2) a nd (4,2). Notic e the tw ose micircles are dra wn antic lockw is e (CC W).

Applying the transformation yields boundary wideline se gments from (4,2) to (12,14) a nd from (-4,-2)to (4,10). Should you sketc h the result of c onnec tingthe pos t-tra ns forma tion end points from (4,2) to(-4,-2) w ith a n origin c entered se mic ircle havingra dius 2√5 and the othe r two endpoints from (4,10)to (12,14) w ith a c omparable circle c entered at(8,12) you c an obse rve the effec t is ae sthetica llyla cking. With dot-da sh line styles, it is important toke ep the prope r s ense of dire ctional movementcons istent w ith the effe ct of a trans forma tion. N otice he re that the two s emicircles would now properly bedraw n c lockw is e (CW ). Line style c an ca use a view er to be a ware of the direction in which a line is draw n. Attention must be pa id to s ens e of movementafte r tra nsformation.

A more consistent visual effe ct is to transform theappropria te se mic ircle from x2 + y2 -4 = 0 into the appropria te portion of the ellipse 10u2 -16uv +8v2 -64 = 0 for the transformed 'leftmost' e nd ca p to close an e nd of the wide line. With ge ometric transformation of graphic e ntities , a ques tionimpleme nters must a nsw er is w hat s hape is a dot orda sh in a line invoking line style as a n a ttribute.

If the graphics system is to provide inherentsupport for such attributes as width of a line ordot-dash style of a line as a geometric attribute,then reproduction of effects such as would be thecase for stretching a rubber sheet or a 150% copierenlargement can guide our decision regarding

[ ] [ ]���

�

�

���

�

�

1fe

0db

0ca

1yx = 1vu

geometric attributes. The rubber sheet ormagnifying glass model for functionalityencourages a geometric interpretation of theattributes line width and line style.

By contrast, consider a different physical model tobe the motivating reference for a computergraphics functional capability. Architectural andengineering drawings have a number of standardpaper sizes. An 'A' size and an 'E' size drawing canbe produced for the same picture content.Typically a draftsman will use the same mix ofpen nib sizes in both drawings. Within anydrawing, a draftsman likely will use severaldifferent pen nibs for line width to representdifferent meanings. For example, a map will usevery thin lines for unimproved, secondary roadsand will use much thicker lines for majorsuperhighways or a dual carriageway.

A quill pen writing instrument can provide ananalogy for cosmetic line width. Suppose adraftsman keeps a collection of ostrich, turkey,chicken, pigeon, sparrow, and hummingbirdfeathers. The choice of feather quill <pen nib> thusdetermines line width <weight>.

Pa tterns within a n are a may have s ymbolic me aning.Civil e ngine ering a pplic ations nee d to repre sentconc rete and othe r materials. Line pa tterns are onewa y of re pre se nting the material. If such a dra wingis replic ate d on la rge r siz e paper, the line sre prese nting the pa tte rn for concrete s hould not havewidth e nlarged. The effe ct to be reproduce d by acompute r gra phics package is provision of ase le ction of line w idths for cos me tic visual effe ct. Inthis situation line width a nd end cap s hapes and dot-da shes should not be s ubjec t to trans forma tion pe rse . The multiple pe n nib size mode l for func tiona lityties a cosme tic interpre tation to the a ttribute s LineWidth a nd Line Style.

From my deve lopme nt da ys as a graphic s sys te mimpleme nter, I'll mention in pas sing that line style ne eds c areful attention to consistenc y. Even indraw ing a full circ le, c are must be taken to ha ve thestarting point on the circumfere nc e c ha nge s o tha trota ted full c irc le s in a pse udo-a nimation s equenceha ve cosmetic line style properly rotated.

5. Clipping

Clipping is another ta sk to w hic h multiple inte rpretations c an be a pplie d. Ma ny te xtbooks de alonly with ma thema tical c lipping, that is, traditiona lge ometric inte rse ction c alc ulations. In traditional

ge ometry, pa ra lle l lines ne ve r mee t or sha re acommon point. Non-para llel lines do inters ec t a ndshare one and only one c ommon point. Such is notthe cas e in ra ste r gra phics . Parallel line s can s harecommon pixels and non-pa ralle l lines ca n inters ec tin more than one common pixel in a ra sterenvironme nt. The line se gme nt from (1,1) to (6,4) is pa ra lle l to the line s egment from (0,0) to (5,3) yettheir ras tered re prese ntations s ha re three pixe ls incommon. The line se gme nt from (0,0) to (5,4) andthe line segme nt from (0,0) to (5,3) interse ctge ometric ally only at the origin yet their rasteredre prese ntations s ha re three pixe ls in c ommon.

With ra ster graphic s, it ma ke s a notice able, visible differe nc e w he the r one a nalytica lly c lips a linese gment locus the n ras te rs the s horte ne d linese gment or w he the r one just s tra ightforwardlyra sters the full line se gme nt onto a much la rge rca nvas then cuts out the smaller 'clip window' ofinte res t <sci ssor cli pping>. I consider a pre-ras te ring clipto be a n algebraic or ma the ma tic al clip or, tha t is, a traditional ge ome tric clip. I cons ide r a pos t-ras terclip to be a s cis sor c lip.

Ra ster pixels, conc eptua lly, are s mall, ze ro-area dots ce ntere d at the intege r grid intersec tion points <l at tice

points> of a rec tangular, Carte sia n coordina te systemin w hic h both the vertic al and horizontal grid linespac ing betw ee n pixels is unity. A clip border willbe a line se gment s pec ified by its be ginning andending pixel c oordinates . Pixels falling e xa ctly on aborder line se gme nt are, by definition here, withinthe pic ture.

Analytical c lipping e xa mines the mathe matic allocus of eac h gra phic entity without re gard to itsattribute s. Any portion of an infinitely thin lineouts ide the ac tive clip window is dis ca rde d. Onlythat portion of the idea liz ed line se gment, now w ithne w, like ly fractional, end points is reta ined. If a nyportion of the line se gment does lie within the c lipwindow, then a ttribute s suc h as width a nd style a reapplied. Whe n a line s egment's mathematica l loc us lies wholly outside the clip window, it is disc arded inits entirety a nd no portion of its wide nedre prese ntation is draw n eve n if the w idene d portionwould fall ins ide the clip window.

Cosmetic clipping c an be thought of as firstre ndering an e ntire picture on a n infinite a rea c anvas then subs equently using a pair of sciss ors to c ut outthe clipped window of interes t. I think of s cis sorclipping as 'raster firs t the n c lip'. I think of ana lytic clipping as 'mathematica lly c lip firs t’ then ra ster.

Suppose a re ctangular clip area is spec ified to have side s parallel to the axes. Let the a re a be, borderinclusive , define d by the a re a s urrounded by the fourline s y=0, x=10, y= 6, x= 1. Se ven-pixe l w ide linesha ving three pixe ls on e ach s ide of the re spectivere fe rence line se gments could be the line se gme ntfrom (9,0) to (9,8) {the line x= 9} and the line se gment from (0,7) to (10,7) {the line y=7}.Analytic clipping w ould rejec t c omple te ly the liney= 7 while the line from (9,0) to (9,8) would be re vised to a s horte r line s egment from (9,0) to (9,6).Sc is sor c lipping would reta in both line s forconc eptua l ras tering follow ed by a sc is sor c lip.

As sume an identity matrix tra nsformation. Analyticclipping would display for the line x=9 the 35 pixelsinside the c lip a re a. The line y=7 would c ontributeze ro pixe ls.

Sc is sor c lipping would displa y 50 pixels derive dfrom ea ch of the tw o original line s x=9 and y=7.While the re fe rence line y= 7 per s e a nd its 'uppe rwide pixe ls' a nd 'leftmost pixels' would be dis ca rde d,y= 7 would contribute 30 pixels bounde d by there ctangle formed from lines y=4, x=10, y=6 a ndx= 1. The line x=9 w ould nee d to contribute its 35pixe ls within the c lip a rea . The resulting picture isthat whic h w ould come from ra ste ring the full linesthen us ing a pair of s cissors to c ut out the clip area from the pos t-ras te red picture. With sc iss or clipping,a la rge picture c ould be ge ne rated in s ucc es sivesw athes . For a ny form of ra ster monitor or dot-ma trix printer us ing a bit ma p ima ge and havinglimited memory, a picture dra wn as succ ess ivesw athes s hould be identical to the sa me picture draw n from a s ingle ra stering of the full picture intoa ve ry la rge memory.

For future refere nc e, I'll note he re that I'vecons cious ly double counted in individua l Scissorclipping line segme nts the pixels in a rec ta nglebounded, edge inc lusive, by y=4, x=10, y=6 a ndx= 6. In a re place me nt pixel draw ing mode it won'taffe ct the final picture se en on a display. In an‘Exc lus ive O R’ draw ing mode , wha t should happento doubly draw n pixels ?

Now let's cons ide r some spe cific rastering s ituationswhic h c an aris e. I'll contrive my exa mple to offe rinte ger inte rs ections of line s a nd shall not cove radditiona l c onsiderations a rising from non-inte gerinte rse ctions. Le t the lowe r clip boundary be y=1and the rightmost bounda ry be x= 9. The leftmostclip boundary can be x=0 and the uppe r boundaryca n be y= 3. For refere nc e here, I'll claim a line se gment from (0,0) to (27,3) will ras te r to:

5 pixels on sc anline y=0 from x= 00 to 049 pixels on sc anline y=1 from x= 05 to 139 pixels on sc anline y=2 from x= 14 to 225 pixels on sc anline y=3 from x= 23 to 27

The line segme nt from (0,0) to (27,3) inte rs ects the boundary line y=1 a t (9,1). It a ls o inters ec ts theboundary line x=9 a t (9,1). M athematica linte rse ction leaves a single point. Eve n a s ingle width, sc iss ored line should dra w the 5 pixe ls at(5,1) (6,1) (7,1) (8,1) and (9,1); they clea rly a rewithin the bounda ry inclusive clip area . W ith w ideline s, some se nse of s lope would be nee ded tode te rmine how to do wide ning for this a nalytic clipinstanc e of a single point (9,1)..

To c ontinue the e xa mple, suppose the rightmost clipboundary were x=8. The line s egment from (0,0) to(27,3) ma the ma tic ally ha s its theoretic al locus wholly outside the clip are a.. Scissor clipping w ouldpres ent the four pixels (5,1) (6,1) (7,1) (8,1).

Ne xt le t the c lip bounda rie s be given by y=0, x=23,y= 3 and x=4. A te xtbook clip would produce the linese gment from (4,4/9) to (23,23/9). Rounding tora ster a perturbe d line segme nt from (4,0) to (23,3) isnot particularly ac curate.

Ac tually ras te ring a line w ith a repe titive algorithmis not ne ces sa ry to de te rmine starting or te rmina tingrun lengths. The pixels too c an be simply a direc tma thema tical e valua tion or a run length slic eevaluation. Cogniza nce a lso must be taken of theboundary line and its cross ing direction. In the above example , in which a n 'integer' bounda ry line y= K isexamine d for inte rs ection w ith a n x-ma jor liney= y0+((y1-y0)/(x1-x0))(x-x0), one ca n use y= K-_ toca lc ula te one extre me of the run a long the s canliney= K and use y= K+_ to c alculate the othe r e nd of therun along the low er boundary. Clipping for rasterde vices a lso w ill need to proces s consiste ntly equalerror mea sure ins ta nce s suc h as de scribed byBoothroyd and Hamilton.

6. Drawing Mode

Draw ing mode s <such as Replace, Or, And, Exclusi ve Or , ...>

are not uncommon as a compute r gra phics attribute .The que stion for a use r can be: OK , I know w hat thedichotomous logic operation A ND is ; w ha t a re theoperands for this bina ry operation on pixe ls ? D oe sdraw ing mode a ffe ct a graphic s e ntity s uch a s a line as a whole or is it a combina toria l s pe cific ationme re ly for pixels or is it me ant to a ffect only bits within an impleme ntation-spec ific, positionde pe nde nt, ordere d bit s equence interna l

re prese ntation of a pixe l before c olor table expa nsion? W ha t's c orrec t depends upon one 's re fe rence mode l.

For simplicity, I shall restrict my brief discussionof drawing modes to a monochrome or bi-levelraster display environment in which a bit plane isused to refresh a display device. Pixels can be ONor OFF.

Exclusive Or (EXOR) is a drawing mode whichusually can be relied upon to illustrate differentpicture presentation depending upon one's choiceof reference model. It also is a good test to verifyquirks in an algorithm; incautious use of 8-waysymmetry in an incremental circle generation isexposed when drawing in EXOR mode if octantend points are doubly dotted.

In a Polyline from (0,0) to (3,0) to (7,0), howmany times is the pixel at (3,0) drawn? If it is boththe last pixel in the first line and the first pixel ofthe second line, it disappears owing to beingdrawn twice. A convention such as the first linesegment in a polyline owns both its first and lastpixel while all other line segments do not owntheir first pixel would cause a single drawing ofthe pixel at (3,0). What would it do for contiguouslines used to outline a closed simple area? Wouldpixels still be 'double dotted'?

Joined line se gme nts w ith a s mall angle be tw eenthem ofte n s ha re ma ny pixels in common. Ra ster thepolyline from (0,0) to (10,0) to (0,1). What should be expe cte d using an EXOR draw ing mode? Theconvention des cribe d in the prec eding para graphcould lea ve the pixel at (10,0) visible but rathe rlone some as its immediate neighbors w ould ha vedisa ppe ared ow ing to double dotting.

7. Conclusion

At the re nde ring le vel a nd for the display proc es singpipe line it is ke y to de fine cle arly what objec tive orview able action is intended. Parts of a graphic ssystem ca n not be s pec ified in a vacuum. If are fe rence mode l is establis he d first, then decisions re ga rding impleme ntation de ta ils c an be ma de more cons istently.

8. Acknowledgements

Thanks are owed to my colleagues on the IBM3270 PC/GX graphics workstation at thedevelopment laboratory in Hursley, England,especially to Nancy Bull, Mike Davis, Adrian Gay,

Brian Middleton, Bryan Roberts, and Norm Sheen.I appreciate the support and encouragement fromSteve Cunningham, Joaquim Madeira, and MikeMcGrath for this GVE workshop. The specificnumerical examples here were a part of an invitedpresentation at Ausgraph’90 InternationalConference on Computer Graphics; MichaelGigante did a great job with the conference.

References

Arnold, D.B. and Bono, P.R. CGM and CGI:Metafile and interface standards for computergraphics. Springer-Verlag. 1988.

Boothroyd, J. and Hamilton, P.A. Exactlyreversible plotter paths. Australian ComputerJournal 2(No.1) p20-21. 1970.

Bresenham, J. E. Pixel processing fundamentals.IEEE Computer Graphics and Applications,16(No15), p74-82. (January, 1996)

Bresenham, J. E. Computer graphics attributes andreference model alternatives. Proceedings:Ausgraph’90 International Conference onComputer Graphics. p413-424. (September, 1990)

Bresenham, J. E. Ambiguities in incremental linerastering. IEEE Computer Graphics andApplications, 7(No.5), p31-43. (May , 1987)

Gay, A. C. Experience in practical implementationof boundary-defined area fill. Fundamentalalgorithms for Computer Graphics, edited by R. A.Earnshaw. Springer-Verlag. p153-160. (1985)

Henderson, L.R. and Mumford, A.M. Thecomputer graphics metafile. Butterworths. 1990.

Posch, K. C. and Fellner, W. D. The circle-brushalgorithm. Theoretical Foundations of ComputerGraphics and CAD, edited by R. A. Earnshaw.Springer-Verlag. p857-878. (1988)

Appendix

Strengthening Visual Skills by Recognising Rendering Algorithms

Rosalee WolfeDePaul UniversityChicago, Illinoiswolfe@cs.depaul.edu

Abstract

Computer science students contemplating a career ingraphics need to develop a visual sense, but traditionalcourse topics do not meet this need. Visual analysis isa teaching technique developed for computer scienceinstructors that helps impart this ability. Through theuse of a few visual cues, students learn to visuallyidentify surface algorithms, shaders and lightingtechniques. An interactive software package calledTERA (Tool for Exploring Rendering Algorithms)provides nearly a million image combinations thatstudents can use to practice their visual identificationskills.Keywords: algorithm visualisation, computer graphicseducation.

1. Background

For a successful career in computer graphics, computerscience majors need to develop a visual sense inaddition to their technical knowledge. Computerscience graduates hired by the graphics industry willwork side-by-side with artists, and it is essential foreffective communication that they have an appreciationand enthusiasm for the visual aspects of the field. Inaddition to programming skills and knowledge ofoperating systems, recent labor market analyses list“strong visual style sense” and “understanding howartistic elements work with technical elements,” asdesirable traits for programmers and technical directors[1,2]. An increasing number of graphics houses wantto see a demo reel or a portfolio from computer sciencegraduates who count Unix shell scripting, C++programming and the mathematical derivation ofNURBS in their skill set.

There are many exemplary resources [3, 4] andlearning tools that allow students to examine aspects ofgeometry [5, 6, 7, 8, 9, 10], rasterisation [6, 11],illumination [6, 12, 13, 14] and the entire 3Drendering pipeline [13]. These present visualisationsthat a student can use to explore and study an algorithmin isolation. This is a sound pedagogical approach fora first introduction, but eventually a student will needto understand an algorithm’s behavior in co-operation

with other algorithms and in the context of a finishedimage.

Unlike other specialities in computer science, thechoice of a graphics algorithm usually cannot be basedsolely upon space, speed and implementationconsiderations. Equal in importance to time andmemory requirements is the visual effect that agraphics algorithm produces. In fact, visualappearance will be the overriding factor for someapplications. Visual understanding is essential todeveloping and debugging new algorithms. Forexample, when writing a shader, a programmer looks ata rendered sample, analyses how its appearance differsfrom the ideal and uses the information thus gained torefine the shader. An essential aspect to developing avisual sensibility is the ability to identify and comparerendering algorithms. Visual knowledge can preparestudents to answer such real-life questions as,“Algorithm X gives us exactly the effect we want, butit is prohibitively expensive. Can we get a similareffect by tweaking our implementation of AlgorithmY?”

Although the graphics industry desires a developedvisual sensibility in new computer science graduates,there is no place in the traditional curriculum wherestudents can gain this knowledge. The discipline ofcomputer science draws on mathematics andemphasises algorithm study and development, whichare almost entirely text-based and involve littlevisualisation. Computer science instructors often viewthe prospect of discussing the visual aspects ofcomputer graphics as a daunting, if not overwhelmingendeavor. Typical reactions include

I’m not an artist! How am I supposed to gainthe requisite background to teach this?

I don’t have time for this in class – I have toomany algorithms to cover as it is.

Both reactions are understandable and justifiable.Computer science instructors are not artists and theirgoal does not include producing artists. Further, thereis an enormous amount of technical knowledge in thecomputer graphics discipline, and attempting to decide

what topics belong in an introductory course has beenthe topic of many papers. Any approach to theproblem of inculcating a visual sensibility has to beone that does not take up much lecture time andconstitutes a teaching method that computer scienceinstructors are willing to try. This paper proposes thatvisual analysis is one approach that fits theseconstraints.2. Visual Analysis

Visual analysis is a teaching technique developedspecifically for computer science instructors thatimparts this knowledge [15]. The technique stems fromcritical analysis, which establishes a structure for

examining works of art [16]. Students in the visual artslearn to describe and compare works in terms ofdesign, concept and media.

Instead of design, concept and media, visual analysisteaches students to recognise a small number of visualcues, including:

• visibility of polygon faces or polygon edges• transition from light to dark on diffuse-reflecting

surfaces• color and shape of specular highlights• presence of transparency, reflection, refraction,

patterns and textures

1. First, determine the surface algorithm.Appearance Algorithm

Outlined polygons. Object’s far sides are visible. Uninterrupted horizonlines.

Wireframe

Outlined polygons. An object’s far sides are not visible. Occluded objectsare invisible.

Hidden-line removal

Filled-in polygons. Presence of refraction in transparent objects. (Presenceof reflection, shadows are also helpful)

Ray tracing

Filled-in polygons. No refraction, reflection or shadows. Z-buffer

2. Determine the shader.For Z-buffer

One color per object. Objects appear flat, as if cut from paper. ConstantOne color per polygon. Objects now appear to have a shaded contour. FacetedSmooth transitions from light to dark. Specular highlights follow polygon

edges.Gouraud

Smooth transitions from light to dark. Specular highlights are white,compact and elliptical. “Shiny plastic” look.

Phong

For ray tracingOpaque, colored surface. No highlight. DiffuseOpaque, colored surface with highlight. Diffuse and SpecularTransparent object that appears to bend light. TransparencyPart of scene is visible in the surface of the object. Reflection

3. Determine additional surface interest (both z-buffer and ray tracing)Image appears to have been pasted or glued onto object. 2-D texture mappingObject appears to have been carved from a solid substance like stone or

wood.3-D texture mapping

Object surface appears rough or wrinkled, but its profile is smooth. Bump mapping

4. Light sources Harsh shadow / lighter shadow. Low/high ambient Sharply defined shadow. Point light Presence of a penumbra, but degree of shade is constant within penumbra. Area light Color bleeding; deepening of shadow in corners, under furniture. Radiosity

Table 1: Visual cues and algorithms(visual equivalencies omitted).

• sharpness of shadow• interactions between adjacent diffuse reflectors

(color bleeding)

These cues are usually sufficient to identify a renderingalgorithm as Table 1 demonstrates. Informal studieshave shown that both instructors and students find thislist of cues non-threatening and easy to spot in animage. By learning to observe and describe thesecues, students are able to identify rendering algorithms.

3. Classroom Presentation

In the first lecture, the teacher discusses imagesportraying three or four algorithms that produce starklydifferent effects. For example, the four surfacealgorithms of wireframe, hidden-line, ray tracing andz-buffer create distinctive effects and can bedistinguished by the use of only three visual cues. SeeFigure 1. Each week, the teacher adds more algorithms,and by the midterm, the class has examined allcommonly used rendering algorithms. Until themidterm, the teacher emphasises the characteristicvisual behavior of the algorithms. After the midterm,the teacher shows how multiple algorithms can achieveequivalent effects. By the end of the course, studentslearn to identify the surface algorithm, shaders andtypes of light sources.

While discussing a visual cue, the teacher presents twoor three images that demonstrate it. After explainingthe visual cue, the teacher gives the name of thealgorithm that creates the cue. The teacher then showsa short series of images and invites students to identifythe visual cues and then suggest a rendering algorithm.Students should specify the cues first so that they learnto spot them in the context of an image. Anypremature guesses as to the identity of the renderingalgorithm are met with the response, “Cues first!”.

After the first class meeting, discussions on visualanalysis begin with a short series of images that reviewthe rendering algorithms from the previous meeting.Students first name the visual cues and then suggest arendering algorithm.

The methodology requires only five minutes of anhour-long lecture. Such a small amount of time willnot significantly impact the presentation of othertopics, especially if the teacher covers visual analysisduring the last five minutes, when student focus isbeginning to wander. In fact, there is a way ofpresenting this material without using any lecture timeat all, as will be discussed in the next section.

4. Study Materials

Students need to practice visual analysis outside of theclassroom. While posting images on the Web orpointing out examples in textbooks is a start, studentssay that they benefit more from the question-and-answer sessions held at the end of a lecture. Simplylooking at images does not promote active learning.

Having students learn a rendering package helps a bitbecause students can choose parameters and view theresulting images. However it takes a significantamount of time to learn a package, and since thestudents have a priori knowledge of the surface andshading algorithms, this approach does not provide astudent with a means of self assessment. A betterapproach is to provide students with an easy-to-useinteractive tool that can demonstrate any rendering orshading algorithm while encouraging active learning.

TERA (Tool for Exploring Rendering Algorithms) isan interactive program that facilitates comparativestudy of the visual effects of rendering algorithms [17].See Figure 2 for an annotated screen dump of TERA.

polygons

opaquerefraction,reflection,

shadow

outlinedfilled-in

ray tracing z-buffer hidden-line wireframe

Y Y NN

Figure 1: Three visual cues identify four surface algorithms.

A student can choose a scene and select a renderingalgorithm for any object in the scene. Students canpractice visual analysis using TERA. In “Self Quiz”mode, students select a scene, and TERA presents itwith each object rendered by a random algorithm.Students then guess the rendering algorithm for eachobject in the scene. TERA responds with “Correct”,“Try Again” or “Close Enough”.

The “Close Enough” response is for those cases wheremultiple algorithms produce similar visual effects. Forinstance, Gouraud and Phong shading produce similareffects when no highlight is present.

Always available is the “Tell me more” button, whichprovides specific feedback about a student’s lastalgorithm selection. When a student is in “Explore”mode, the “Tell Me More” button will activate a popupwindow describing the relevant visual cues. In “SelfQuiz” mode, when students get a response of “CloseEnough”, they can click on “Tell Me More” for anexplanation. A pop-up window will list the algorithmthey picked, the actual algorithm and a specificexplanation of why the two algorithms producedeffects that are visually equivalent. For example, if aflat surface is very evenly lit, then Gouraud shadingmay produce variations in shade that are so subtle thatthe result looks like constant shading. In this situation,

if a student picks “constant” and receives the “CloseEnough” response, the “Tell Me More” button willactivate a pop-up window containing the relevantexplanation. See Figure 3.

The new version of TERA is capable of creating nearlya million images for students to analyse. The imagescover surface algorithms, z-buffer shaders, ray tracingshaders, texture mapping, bump mapping, lighting andvisual equivalencies.

Some teachers do not discuss visual analysis in theclassroom, but give a short demonstration of TERAand tell students that TERA is available in the lab.TERA has enough appeal that students are drawn to it,and they spend enough time with it that they can scorereasonably well on a visual identification test [18].

5. Results and Feedback

In my experience, visual analysis adds excitement anda sense of the “big picture” to introductory graphicscourses. Students may not be able to implement everyrendering algorithm when they leave the course, butthey will be able to recognise them and know theirnames, which provides a starting point for furtherinvestigation.

Figure 2: A sample TERA session.

Selectedobject

Algorithmmenu

ImageSelectionmenu

A visual sensibility, fostered by visual analysis,enhances a depth of knowledge of those renderingalgorithms that students do implement. Because theyare already familiar with the visual behavior of thealgorithm, they have expectations of how an imageshould appear and can perform more of theirdebugging on their own without outside help. Studentsno longer ask, “Is it right?” They state, “There’ssomething wrong with my highlight.”

A very exciting development is the favorable feedbackfrom digital artists. This approach promises to help artstudents “grasp the differences between light models,surface algorithms and shaders” [19]. Further workincludes the exploration of developing the approach toexpand the possibilities of better communicationbetween digital artists and computer scientists.

References

[1] Regan and Associates. A Labor MarketAnalysis for the interactive Digital MediaIndustry: Opportunities in Multimedia.Sunnyvale, CA: North Valley PrivateIndustry Council (NOVA), 1997.

[2] Public Affairs Coalition of the Alliance ofMotion Picture and Television Producers,Making Digits Dance: Visual Effects andAnimation Careers in the EntertainmentIndustry. Sunnyvale, CA: NOVA PrivateIndustry Council, 1997.

[3] Owen, G. Using Hypermedia to TeachComputer Graphics. Presented at theGraphics and Visualization Workshop inEurographics ’93 Barcelona, September,1993.

[4] G. S. Owen. Integrating World Wide Web

Figure 3: TERA in “Self Quiz” mode.

Technology into Courses in ComputerGraphics and Scientific Visualization.Computer Graphics 29(3) pages 12-14,August 1995.

[5] D. Schweitzer. Designing InteractiveVisualization Tools for the GraphicsClassroom. In Proceedings of the Twenty-third SIGCSE Technical Symposium onComputer Science Education. pages 299-303, 1992.

[6] P. Min.www.cs.princeton.edu/~min/cs426/applets.html

[7] A. Shabo, M. Guzdial and J. Stasko.Addressing Student Problems in LearningComputer Graphics. Computer Graphics29 (3) pages 38-40, August 1996.

[8] R. Klein, F. Hanisch, W. Strasser. Web-Based Teaching of Computer Graphics:Concepts and Realization of an InteractiveOnline Course. Computer Graphics Annualconference Series, Conference Abstracts andApplications. pages 88-93, 1998.

[9] Y. Zhao, J. Lowther and C. Shene. A Toolfor Teaching Curve Design. In Proceedingsof the Twenty-ninth SIGCSE TechnicalSymposium on Computer Science Education.pages 97-101, 1998.

[10] Y. Zhou, Y. Zhao, J. Lowther and C. Shene.Teaching Surface Design Made Easy. InProceedings of the Thirtieth SIGCSETechnical Symposium on Computer ScienceEducation. pages 222-226, 1999.

[11] A. C. Naiman. Interactive TeachingModules for Computer Graphics. ComputerGraphics 30 (3) pages 33-35, August 1996.

[12] Ansari, M. Simple IlluminationVisualization. Department of Computer

Science, DePaul University, Chicago, 1993.[13] E. Wernert. A Unified Environment for

Presenting, Developing and AnalyzingGraphics Algorithms. Computer Graphics31 (3) pages 26-28, August 1997.

[14] D. Gould, R. Simpson and A. van Dam.Granularity in the Design of InteractiveIllustrations. In Proceedings of the ThirtiethSIGCSE Technical Symposium on ComputerScience Education. pages 306-310, 1999.

[15] A. Sears and R. Wolfe, Visual Analysis:Adding breadth to a computer graphicscourse. In Proceedings of the Twenty-sixthSIGCSE Technical Symposium of ComputerScience Education. pages 195-198, March1995.

[16] R. Arnheim. Art and Visual Perception: APsychology of the Creative Eye. 2nd ed.Berkeley: University of California Press,1974.

[17] R. Wolfe, A. Sears. TERA: an interactiveTool for Exploring Rendering Algorithms.Journal of Computing in Small Colleges 10(4) pages 41-46, February 1995.

[18] R. Wolfe, S. Grissom, T. Naps and A. Sears.A Tested Tool for Teaching ComputerGraphics. Journal of Computing in SmallColleges 12 (2) pages 70-77, November1996.

[19] D. Eber. Personal communication,December 1997.

MAVERIK: A Virtua l Reality System for Research and Tea ching

Toby Howard, Roger Hubbold and Alan MurtaAdvanced Interfaces Group, Department of Computer Science

University of Manchester, United Kingdomtoby@cs.man.ac.uk

Abstract

This paper describes some experiences with the useof the MAVERIK system for supportingundergraduate and postgraduate teaching andresearch. MAVERIK is a high-level system forcreating and managing interactive virtualenvironments. It is available free under the GNUGeneral Public Licence. MAVERIK is modular andextensible, and its use of Mesa, the free OpenGL-likegraphics system, means that it can be run on low-cost PCs, making it especially suitable for use incomputer graphics and visualization education. Wediscuss the novel architecture and main features ofMAVERIK, and illustrate its use by presenting casestudies of projects undertaken by our students.Keywords: Computer graphics education,MAVERIK, Mesa, Virtual Reality.

1. An Introduction to MAVERIK

MAVERIK is a system for managing graphics andinteraction in Virtual Reality applications [1, 2]. It isspecifically designed to address the challenges ofhighly interactive virtual environments containingmany objects with complex geometry. MAVERIKruns on PCs under the GNU/Linux operating system,using the free Mesa OpenGL-like graphics API [3],and can take advantage of 3D acceleration hardwareif present. It also runs on Silicon Graphicsworkstations using Irix/OpenGL [4].

MAVERIK is free software, released under the GNUGeneral Public Licence; the distribution includes allof the C source code, for both MAVERIK, and also anumber of example applications (with data). Thissets it apart from most commercial virtual realitysystems, for which the source code is typically notavailable – or may be available, but at a costprohibitive to educators. MAVERIK has a highlymodular structure, and comprises a micro-kerneland a collection of supporting modules. The micro-kernel implements a set of core services, and aframework that applications can use to buildcomplete virtual environments and virtual realityinterfaces. The supporting modules contain default

methods for optimised display managementincluding spatial management, culling, interactionand navigation, and control of conventional and VRinput and output devices. MAVERIK’s structureallows these default methods to be customised tooperate directly on application data, so that optimalrepresentations and algorithms can be employed.

The MAVERIK micro-kernel uses a simple object-oriented class structure. Following normal OOphilosophy, each class has an associated set ofmethods. For example, for geometric primitives,default methods are provided for displaying them,and for selecting them with a mouse. To keep theimplementation simple, methods are implemented ascallback functions. Changing a method is as simpleas writing a new (or modified) callback function andregistering it with the MAVERIK kernel. The kernelalso provides a mechanism for defining new classes,making the system extensible. This approach was adeliberate design decision – we wanted the system tobe easy to understand and install. In a studentcontext, we have found that using C and callbackshas meant that those with basic programming skillscan quickly become proficient at using the system.

A key difference between MAVERIK and manyother VR systems is that MAVERIK does not use itsown data structure for storing application data. Ofcourse, the kernel does have data structures formanaging classes and their methods, but as far aspossible we have tried to avoid imposing datastructures on applications. Instead, classes of objecttypes can be created which suit the needs ofparticular applications. Methods for displaying andinteracting with these objects are then defined andregistered with the kernel. This means that only asingle representation of the application data needsto be maintained. Any changes to this data areautomatically reflected as soon as the next frame isdisplayed, because MAVERIK uses immediate moderendering. The default supporting modules takemuch of the work out of writing the display andinteraction methods.

MAVERIK is distributed with nineteen defaultobject classes, and standard methods for navigationand for picking objects using a standard mouse andkeyboard. In our research laboratory we also employ3D mice based on Polhemus magnetic trackers, andcode for handling these is included in thedistribution. We also use head-mounted displays andlarge-screen projection systems and MAVERIKincludes the code needed to compute stereoprojections, and to synchronise frame buffer updates.Another part of the distribution is a set of tutorialexamples, which are described in the documentation(in PostScript, HTML, and on-line man pageformats), and some more advanced demonstrationswhich illustrate the use of MAVERIK for some ofour own research applications.

2. Using MAVERIK for Teaching

There are several reasons we believe MAVERIK isuseful for teaching computer graphics andvisualization:

1. MAVERIK is free. Students can run it on theirown PCs, allowing them to study in their owntime as well as on campus. Similarly, othereducators have free access to the system.MAVERIK runs on relatively low-cost PChardware, relying only on other software whichis also available free.

2. MAVERIK is a fully featured, professional-level system, not a small subset developedspecifically for teaching. We use it in our ownresearch to explore very large, complex virtualenvironments. For example, polygonalrepresentations of some of our industrial CADmodels would amount to more than 100 millionpolygons. Examples can be found on our Webpages [5].

3. MAVERIK implements a range of algorithms –such as culling, using a hierarchy of boundingvolumes or occlusion techniques, objecttessellation, computing projections, andstereoscopic viewing – which students can studyto see how they work. The availability of sourcecode means they can learn by example and byexperimenting for themselves by changingdefault methods.

4. MAVERIK is distributed with default methodswhich handle the most common cases – such asprimitives like boxes, cones, spheres, cylinders,tori, polygon meshes. These serve as examplesof how to build other types of objects: NURBS

perhaps, or even Jell-O! For example, as wedescribe in one of the case studies, it provedstraightforward to add a new class of object toimplement superquadric shapes.

3. Case Studies

We use MAVERIK extensively in our own teaching–for final year six-month undergraduate projects, andfor postgraduate work – as well as our own research.In this section we present short accounts of aselection of recent and on-going student projectswhich illustrate the diverse range of applications ofMAVERIK in an educational context.

3.1. Builder – Direct 3D Interactive ConstructionBuilder is a 3D object manipulation program (seeFigure 1). It was designed to explore techniques forselecting and manipulating hierarchically structuredobjects in 3D space, using two-handed manipulation.The manipulation is carried out using a pair of 3Dmice, and the resulting images are shown on a stereoprojection screen, a head-mounted display, or anormal monitor. Builder uses voice recognition toissue commands, because both hands use the 3Dmice.

Builder is a very dynamic program. Much as in a 2Ddrawing program, objects can be selected andstretched or squashed, using handles attached to keypoints on their bounding boxes. Objects can becreated, deleted, copied, pulled apart, joinedtogether, and positioned in space. Each operationmay be reinforced using sound cues, via a MIDIinterface. It is possible to enter a virtual environmentand to completely rearrange it using thesetechniques.

Figure 1. Builder: an interactive 3D modeller.

Buttons on the mice differentiate between movingaround the environment and picking up objects. Thetwo mice can be used interchangeably, so that it ispossible to pick up an object in one hand and to carrythe object around, using the other mouse to controlnavigation.

Builder relies on a key feature of MAVERIK – that ituses only the application’s data and a set of callbackfunctions to update the display. There is no separatescene graph or other data structure that needs to besynchronised with changes to the application’s data.It also uses MAVERIK’s culling methods to permitquite large environments to be displayed at fastframe rates, which is essential when using an HMD.

3.2. Virtual LegoAnother approach to interactive model building isillustrated in the Virtual Lego project (see Figure 2).Here we consider the joining and breaking apart ofmodels consisting of a set of bricks. It differs fromBuilder in that constraints are applied to only allowthe joining of shapes in a small number of knownarrangements, and the system applies continuousobject clash detection during the assembly process.Interaction is achieved using a combination of 2Dmouse and keyboard.

Figure 2. Constrained Lego brickmanipulation.

3.3. Professor Dijkstra goes WalkaboutOne of the demonstration programs distributed withMAVERIK is a virtual city (“cityscape”), withbuildings, roads, trees, pavements, grassy areas,monuments, and animated avatars. This projectexplored extensions to the cityscape program toinclude a route planning mechanism. The initialmethod used Dijkstra’s algorithm for computing theminimum cost path between any two places. Toprovide a more efficient implementation, a second

version was implemented based on the Floyd-Walshall algorithm. Both of these algorithms areused to find the minimum cost for traversing a graphstructure, and are standard algorithms taught toComputer Science students. The project used theXforms interface builder [6] to provide a GUI. Aspeech recognition system can also be used toinstruct the program. The project is quite interestingfrom two angles.

First, it is an interactive city guide. Using theinterface, you can ask Professor Dijkstra –represented by an avatar – to guide you around thecity (see Figure 3). For example, you may ask to betaken to the railway station, or the public library. Theprogram builds a graph structure for all of theconnected streets, and you are guided along thepavements, traversing roads at junctions orpedestrian crossings, using the shortest route. Theresult is displayed as an animated walk-through. Atany time, you can stop or move away from yourguide, and he will recognise this and wait for you.

Second, it is a nice illustration of the Dijkstra andFloyd-Walshall algorithms in practice. While theprogram is running, obstacles can be introduced,interactively, to block streets – to simulate roadworks, or perhaps an incident like a fire, for example– and the algorithms then search for the best wayaround the obstacles via alternative streets.

Figure 3. Professor Dijkstra goes walkabout.

This program exploits several features ofMAVERIK. First, it uses occlusion cullingalgorithms to achieve good frame rates, even forlarge cities (e.g. 40 blocks square). Second, thenavigation methods can be employed to wanderaround the city at will. Third, MAVERIK’s defaultavatar class is used to display representations of

actions of people in the city. The avatars areanimated, and walk around the city in real-time.

3.4. Head-Driven NavigationSpecifying scene navigation within interactive 3Denvironments is often based on 2D mousemovements and key-presses, or uses 3D mouse ordataglove gestures. This project considered the useof head motion (tracked using a 3D Polhemussensor) to trigger user movement within a virtualenvironment. This may be useful when both handsare already involved in a manipulation task, forexample. Navigation behaviours were devised whichmap head motion onto view rotations andtranslations. Alternative mappings were tried forboth head-mounted displays and fixed large-screenprojection screens. MAVERIK proved to be a goodplatform for investigations of this kind, since theuser is free to customise navigation functions forspecific applications. User tests were performed tocompare the usability of the head-driven approachwith more traditional hand-driven navigationschemes. These included a timed fly-through of atwisting 3D tube, and the navigation of a slalomcourse.

3.5. Out-of-Body ExperiencesNavigation in virtual environments is a difficult task.One of the main hindrances is the limited field ofview afforded by head-mounted displays andprojection systems, as this is typically muchnarrower than we experience in the real world. It hasbeen described as like looking at the world through apair of toilet rolls! This makes life difficult, becauseit means that one tends to blunder into objects thatare just out of view.

An important aspect of virtual reality is the notion ofpresence – that is, of being present in a virtualenvironment and of being surrounded by it. Thus,when navigating around, one is normally presentedwith a view corresponding directly to that of one’svirtual body – an “in-body” or first person view.However, an alternative way to navigate is to followa few paces behind one’s virtual body – a kind of“out-of-body experience”. In this case, one stillcontrols one’s body, but sees it as a separaterepresentation, or avatar. This type of interface canbe found in some PC games, such as Tomb Raider[7].

The purpose of this project was to investigatewhether navigation is more easily controlled usingan in-body or an out-of-body paradigm. One reason

why the latter might be easier is that it affords amuch wider field of view – you can see obstaclessurrounding your virtual body that would notnormally be within your field of view using the in-body approach. The project entailed devising somesimple virtual environments containing obstacles andthen monitoring the performance of volunteers innegotiating their way along prescribed routes. Thecode was instrumented to measure results, such asthe number of collisions with objects, the length ofthe path followed, and the time taken to complete aroute.

3.6. Visualization of NanotechnologyThis project investigated how computer graphics andVirtual Reality techniques might be used to visualizeand control the construction and activity ofnanomachines. The project involved the simulationof molecular interactions, and the modelling ofcollections of molecules organised into mechanicalstructures (see Figure 4).

This was an ambitious project, and MAVERIKproved especially useful – its provision of high-levelvirtual environment management enabled the studentto concentrate on the application problem, ratherthan the infrastructure needed to manage the virtualenvironment.

Figure 4. Visualizing nanotechnology.

3.7. A Clutterizer for Virtual EnvironmentsA virtual environment is typically constructed bycreating accurate geometrical architecture, andpopulating the scene with specific objects (andtextures) at specific locations. Such environments areusually quite tidy. Real environments, however, arerarely tidy, having accumulated clutter over time.

The purpose of this project was to investigatemethods for automatically “clutterizing'” a virtualenvironment, by semi-automatically adding “junk”geometry and texture.

MAVERIK provides default methods for testing forcollision detection between objects in the virtualenvironment, and this feature facilitated thedevelopment of algorithms for rapidly populating theenvironment with large numbers of “clutter” objects,distributed statistically.

3.8. Interactive City GenerationThis undergraduate project investigated methods forautomatically generating realistic cities (see Figure5). The project implemented a suite of three separateprograms; the first generated a plausible layout ofstreets within the city; the second then populated thestreets with buildings, monuments, parks and otherurban features. The final program was an interactiveviewer, which used MAVERIK for visualization, andnavigation within the scene.

Figure 5. Interactive city generation.

3.9. Line of Sight and Trajectory Analysis forScene of Crime Reconstruction

Working in collaboration with Greater ManchesterPolice, we have constructed a prototype virtualenvironment corresponding to a real scene of crime[8]. Subsequently, this undergraduate projectundertook an investigation of two areas: first,methods for testing “line of sight” scenarios, toaccurately determine which parts of a scene arevisible from certain vantage points; second, forinteractively tracing and modelling bullet trajectorieswithin the scene.

MAVERIK provided the virtual environmentinfrastructure, and its default methods for computingvector-object intersections assisted the modelling ofbullet trajectories, and multiple ricochets from rigidsurfaces. The project implemented two kinds oftrajectory simulation: a simple Newtonian model,and a “point mass” model which takes into accountaerodynamic drag on a bullet. Figure 6 shows asimulated bullet trajectory and ricochet.

Figure 6. Visualization of bullet trajectories.

3.10. Simulating Undersea Viewing ConditionsWe are interested in using VR as a means ofrehearsing assembly and maintenance procedures onlarge industrial structures, such as offshore gas andoil platforms. This project investigated the graphicalportrayal of the kind of environments encounteredduring such undersea operations (see Figure 7). Inparticular, the incorporation of visual phenomenasuch as animated caustics, lighting effects withinparticipating media, limited visibility (fogging) andthe presence of particulate matter in the environmentwere considered. This project used MAVERIK fornavigation, interaction and scene management, withadditional use of Mesa/OpenGL to achieve specificgraphical effects. Real time undersea effects (over 20frames per second) were achieved on a modest PCwith a 3D graphics card.

Figure 7. Shallow undersea view featuringanimated lighting effects.

3.11. Interactive Modelling Using SuperquadricsSuperquadrics are a family of solids derived from theparametric forms of the basic quadric surfaces –ellipsoids, tori and hyperboloids. Superquadrics are,however, much more flexible than standard quadricsand can represent a huge variety of shapes. Thispostgraduate project is investigating the use ofinteractively deformed superquadrics for objectmodelling in a virtual environment (see Figure 8).

This involved adding a new, quite complex, objectclass to MAVERIK. In practice this proved to bestraightforward: the application defines a datastructure for the object, and a minimal number ofmethods – which it registers with MAVERIK – tooperate upon it.

Figure 8. Interactive deformation ofsuperquadrics.

4. Conclusions

We have been using MAVERIK as the basis for ourteaching and project work since 1996, and have beenextremely encouraged by its success. Because thesource code is freely available, students have theopportunity to look inside and see exactly how alarge professional-level software system isstructured, and they can examine the algorithms indetail. And because MAVERIK uses Mesa as itsrendering engine, computer graphics students can gofurther, and see exactly how the rendering pipeline isimplemented. As for project work, becauseMAVERIK manages the complexity of the virtualenvironment, students are able to pursue moreambitious projects than have previously beenrealistic.

It is our hope that sharing our positive experiences ofteaching using MAVERIK will be of benefit to othereducators. We would be delighted if others wouldconsider evaluating MAVERIK as a suitable tool forcomputer graphics and visualization education.

Acknowledgements

We would like to thank our colleagues in theAdvanced Interfaces Group: Jon Cook, SimonGibson, Martin Keates, Steve Pettifer, Adrian West –all of whom have contributed substantially to theMAVERIK system. And we’d especially like tothank our students for their enthusiasm, and forimplementing a series of fun and interesting projectsfor us to report: Osian Ap Garth, Tim Davis, SinemGuven, Chris Kirk, James Pearce, Jamie Pratt,Ahmed Rahali, Chris Redburn, James Sinnott, JohnTaylor and Irina Titovich.

MAVERIK was developed as part of the VRLSAproject, funded by the Engineering and PhysicalSciences Research Council (GR/K99701), withadditional support from our industrial partners,CADCentre Ltd, Brown & Root Ltd, and SharpLaboratories of Europe Ltd.

MAVERIK is available in tar and RPM formats fromhttp://aig.cs.man.ac.uk/systems/Maverik/.

References

[1] Roger Hubbold, Dongbo Xiao and SimonGibson. MAVERIK: The Manchester VirtualEnvironment Interface Kernel. In VirtualEnvironments and Scientific Visualization1996, M. Gobel, J. David, P. Slavik, J.J. van

Wijk (Editors).[2] Jon Cook, Roger Hubbold and Martin Keates.

Virtual Reality for Large-Scale IndustrialApplications. In Future Generation ComputerSystems, 14, pp. 157–166, 1998.

[3] http://www.mesa3d.org.[4] http://www.opengl.org.[5] http://aig.cs.man.ac.uk/systems/Maverik/.[6] http://bragg.phys.uwm.edu/xforms/.[7] http://www.tombraider.com.[8] Alan Murta, Simon Gibson, Toby Howard,

Roger Hubbold and Adrian West. Modellingand Rendering for Scene of CrimeReconstruction: A Case Study. In ProceedingsEurographics UK, pp. 169–173, 1998.

Computer Graphics Curricula in the Visual Arts

Dena Elisabeth EberBowling Green State University

Bowling Green, Ohio, USAdeber@christo.bgsu.edu

Abstract

Computers have recently emerged as commonplaceon the scene in a record number of university artdepartments. This rapid change has created theneed for new curriculum in the computer artdiscipline. In designing these courses I see threeideas that are paramount to any computer graphicscurriculum in the visual arts. First, the courses andcourse work should focus on individual expressionand creativity, second, the instruction should includean awareness of the “wow” factor, and third, thecourses should include a balance of computer art,traditional art, computer science, and academicssuch as theory, art history, and criticism.Keywords: computer graphics curriculum, visualarts, higher education

1. Introduction

Against many odds and tight budgets, computershave recently emerged as commonplace on the scenein a record number of traditional college anduniversity art departments. As part of this rapidchange, artists are not only using computers as anextension of many traditional art practices, but theyare also using them as a medium to create uniquedigital art forms. From imaging, digital painting,and 3D modeling, to interactivity, video, and 3Danimation, graphics technologies give artists thepower to use the computer as an expressive art form,thus forging new educational and aesthetic concernsin curricula for computer graphics educators. Thispaper will address a subset of these concerns in ageneral way with the intention of highlighting areasthat individual programs and researchers can furtherexpand.

1.1. Scope of PaperThis paper will articulate some curricular issuesrelated to teaching computer graphics in the visualarts at the higher education level. The 1990Guidelines for Curricula in Computer Graphics inthe Visual Arts, also know as The DreamCurriculum, addresses similar issues, however somehave changed and others have emerged since itspublication [7]. I will touch on a small number of

the emergent and dated topics. Other areas of artsuch as music and theater may also be affected bythe changes in computer graphics, but these domainsare not covered in this paper. Also of greatimportance but outside the scope of this paper iscomputer graphics in kindergarten through twelfthgrade (K-12). This document is concerned withcurricula for students at the college level who havechosen computer graphics as a field of study and as acareer in the visual arts.

Curriculum for computer graphics is quite broadwith applications ranging from graphics and textiledesign to studio art. This paper is modeled after astudio art approach, however, it also encompassesthe commercial arts such as graphics design, interiordesign, textile design, architecture, interior,industrial and packaging design. Both the studioartist and the commercial artist focus on expressionthrough visual media, however, the key difference isthe application of that focus. In graphics design andcommercial arts, the focus is on a client or acustomer and the goal is to use personal creativity tovisually express, design, or illustrate a product oridea. In studio art, the focus is most often on theartist’s personal expression, creativity, or ideathrough visual media.

The suggestions in this paper do not go into enoughdetail to differentiate the nuances between these twogeneral approaches, especially as the line betweenthem becomes more and more indistinguishable.However, it is my intention to highlight that I amwriting from a studio art frame of reference and toleave the application of my suggestions up to thereader.

1.2. What is Art?

Art is a human activity consisting in this,that one man [or woman] consciously bymeans of certain external signs, hands onto others feelings he [or she] has livedthrough, and that others are infected bythese feelings and also experience them.

Tolstoy, from What is Art?, 1897

A work of art, according to Tolstoy [8], is sincere,and it transmits information, feelings, andexperiences through lines, colors, sound, or words.Actualization of the elements embedded in theimagery start with the creator, the creative process,and the ability to see and to know through thatseeing. The work may take varied forms, but to beart, the object, idea, or presentation goes beyond thephysical and contains some aspect of humanexperience. Art may be created with any tool, aslong as the artist can use that tool to put some oftheir true self into the art.

A common approach to teaching studio art is tostress expression of the individual, a focus that goesfar beyond the tool or medium. Artists have alwaysused technology and many have learned to expressthemselves in a way that emphasizes the message,not the medium. There are many other dissentingviews of media, some of which are based onMarshall McLuhan’s idea of electronic media, thatthe medium is the message [6]. Bits that underliethe varied digital representations are all the same,thus setting expressive boundaries upon artists [4].

The balancing of the message and the medium indigital technologies is similar to a problem thatartists have always had when working with newmedia of the time, however, digital technologies adda new twist. Computer hardware and softwaretechnologies are being replaced at the rate of everyeighteen months, if not faster. Rising beyond themedium takes time, but an ever-changing paintbrushthat is highly technological hinders the time artistshave with it. New tools require them to re-learn askill set, which in turn, forces them to constantlyshift between artistic expression and skillacquisition, two different ways of thinking. Teachingcomputer graphics in an art setting demands findinga delicate equilibrium between artistic expressionand technological proficiency, something thecurriculum must address.

2. A General Approach

2.1. Philosophical underpinningsNo matter the focus of a digital art program, be it aspecialization within a traditional university artdepartment, an extension of an established studioarea, a commercial major, or other related areas, thecurriculum must balance digital technology withother concerns. I see three ideas that are paramountto any computer graphics curriculum in the visualarts:

1) The courses and course work should focus onindividual expression and creativity,

2) The instruction should include an awareness ofthe “wow” factor

3) The courses should include a balance ofcomputer art, traditional art, computer science,and academics such as theory, art history, andcriticism.

For students to make expressive art, they must payattention to the message it portrays. This means thatinstructors need to make a pedagogical shift fromtechnique oriented curricular goals to exercises thatlimit the tool set and leave room for experimentation,play, and theme based works. The tool set should beexpanded only after students address content andshow a level of comfort with the medium. In thisway the digital techniques will be one importantconcern, among others. Further, the students thenhave time to figure out what they want to express,thus shaping the medium rather than the mediumshaping them. The critiques and assignments shouldfocus around an idea, which stimulates self reflectionand critical thinking. The premise of this approachis if an instructor stimulates critical reflection anddialogue within and between the students, then theywill have a better understanding of how to shapedigital media to express their ideas. This will in turndramatically improve the outcome of their work [3].

Related to individual expression is an awareness ofthe “wow” factor, a condition that is especiallyabundant when working with digital media. Thischaracteristic takes many forms, but in most cases itis a desire on the part of the student to create workssolely for the application of a cliché or oftenoverused special effect. Not only does this kind ofwork look canned or have an “off-the shelf”aesthetic, but it lacks personal expression andcreativity and shifts the focus away from criticalreflection and back to surface level technique. It isfor this reason that instructors must point out the“wow” factor and de-emphasize it.

Beyond cliché and content, it is my view that thecurriculum should include art theory, criticism, andhistory at least within the studio courses, but mostpreferably in addition to them. Not only does thishelp artists identify their work with other art incomputer graphics and traditional media, but it helpsthem critique it and understand how their ideas canexpand on those of others. Schools such as KeelUniversity stress a 50/50 split between studio andtheory, which provides the student with discourseand data to situate their work in the digital and artworld [2].

Keel University also requires the computer graphicsstudent to work in another studio area such asphotography, print, painting, or sculpture. Thisallows students to see different approaches toexpression that computer graphics can extend. It ismy view that this approach also helps the studentunderstand and contextualize their work with digitalmedia, a new and multifaceted medium.

Not only should the curriculum address computerart, other studio art areas, and theory, but it shouldalso introduce the student to programming. TheDream Curriculum [7] points out that teaching artistshow to program is a controversial notion. At thetime it was written, artists who wanted to use threedimensional (3D) imagery and other image effects intheir work had to write many of their own programs.Since then, powerful image processing and 3Dmodeling packages have been written and the artistno longer has to, but often still does, program.

Also since The Dream Curriculum, use of the worldwide web (WWW) has emerged and many otherkinds of scriptable interactive packages have becomeavailable. There will be times when most artists willneed to write scripts, however what kind, in whatlanguage and for what purpose will prove to bedynamic. Underlying any specific programmingapplication is the basic understanding ofprogramming logic, data structures, functions, inputand output, and variables. At minimum, artistsshould understand these ideas in whatever flavor isavailable, such as C or visual basic. This should be aclass most preferably in computer science becausethe cross disciplinary setting gives both artists andcomputer scientists each a taste of the otherdiscipline and how people in that field think. Thissetting is increasingly important as industry headstowards interdisciplinary art and science teams.

Some artists may choose to go beyond the basicprogramming level and make it a larger part of howthey create art. Some artists and educators believethat programming is one way to break the “pre-set”or “off-the-shelf” aesthetic that makes some workappear unoriginal [4]. It is my view that thisapproach will work for some, but not all.Programming requires a technical, spatial, and linearway of thinking, which some artists employ inaddition to non-linear and emergent approaches.However, there are many art students who do notthink in this way and have talents or intelligences inmany other styles such as personal or bodily-kinesthetic forms [1]. These methods and more canbe integral to making successful works of digital art.

For these students, intuitive software provides thetools they need to make rich and expressive works ofart. There is a lot that focused and creative studentscan do with a pre-programmed digital paint brush.

I believe that the learning objectives for students incomputer graphics and the visual arts should includenot only technical proficiency, but also anunderstanding of the significance of their work andhow they can manipulate the medium to expresstheir message. They can accomplish this through acurriculum balanced in complimentary areas as wellthrough instructors that stress ideas as well astechnique.

2.2. Foundations CoreA common approach to university level art programsis to begin with what is called a foundations core, ora set of classes that address basic art principles.Many programs break this up into the twodimensional (2D) and three dimensional (3D) areas.2D course work often includes drawing,composition, color theory, figure drawing, andpainting, while 3D course work focuses on differentforms of sculpture. Some programs are alsoincorporating sections that cover the concept of time,performance, criticism, theory in contemporary art,interactivity, and installation art. The computer infoundations is threefold, it can be used as aneducational tool to teach concepts such as colortheory, it can be taught as an extension of othermedia such as printmaking or painting, or it can beaddressed as a medium in and of itself.

It is my view that computer graphics as an artmedium should in some way be part of thefoundations core. It is important that students usingall media understand the role that computer graphicsplays in their art making, be it directly or indirectly.In other words, it is meaningful for all artists to havean understanding of digital media as a basic corerequirement, much like drawing is treated as a corefor all art students, which brings up another point ofcontention.

Converse to the argument that non-computer artstudents should be exposed to digital media is theargument that computer art students should takedrawing. Since The Dream Curriculum was written,computer art has become a commonplace major orspecialization. As a result of this change, maycontend that not all the standard core requirementsapply to computer art majors. It is my view thatdrawing and other basic core requirements areabsolutely essential for computer graphics majors.

Recall I stated that art is about seeing and corerequirements, if nothing else, help students learn tosee.

Others say that drawing helps those in computergraphics, a discipline that stresses vertical thinking,to solve problems in a lateral way [5]. Verticalthinking is characterized by a stepwise logicalprocess, while lateral thinking is a non-linearprocess. A vertical approach to finding a solutionmight be to search one entire plane of solutions whenin fact the solution is on another search plane. Thelateral approach will throw the search into a newspace at which time vertical thinking may take over.Most people can think in both ways and according tosome, drawing is one way to exercise lateral thinkingin computer graphics, which is otherwise a verticaldiscipline.

In designing the core curriculum, I think one veryimportant and often overlooked notion, especially forcomputer art majors, is that of time, or 4D. 2D or 3Dcomputer animation, digital video, and interactiveworks of art such as WWW pieces or virtualenvironments (VE) all require an understanding ofhow ideas unfold, change, and take shape over time.Whether it is dealt with in a separate course oraddressed in a section of a universal art core class,the concept of 4D should be included because it isbasic to many computer graphic art forms.

2.3. Enter Computer GraphicsWhether computer graphics is part of the corecurriculum or not, computer art majors must beginwork in their discipline at some time. A point ofcontention is when to start computer art majors incomputer art classes. In some programs, computerart majors begin work with computer graphics at theupper-class level, only after the core classes arecomplete. In a four year program, this means thatthey do not start with computer graphics until theend of their sophomore or the beginning of theirjunior year.

I believe that computer graphics majors in the artsmust start on the computer as soon as they can,preferably in their first year. The first course shouldbe a sampling of the computer art focus areas offeredat the institution. From this course, the ones thatfollow should be a higher competency level in aparticular focus area. See table 1 in section 3 forsome of these areas.

As I have already mentioned, art is a form ofexpression and it is up to the artist to learn how to

shape the medium to do so. This takes time, muchmore than two final years of college can provide.Further, the students need to be aware of thecapabilities that the computer graphics mediumprovides while they form their artistic ideas.

3. Beyond the Core: Focus Areas inComputer Art

The Dream Curriculum mentions that art educatorsneed to decide if computer graphics will be asupport area for other media or a major unto itself.Since 1990, it is more common to find a computerart major or specialization in an art program than acomputer lab which supports other disciplines. It isfor this reason that the following suggestions areintended for computer art majors. I did not indicatecompetency levels or what courses should come inwhat order following the core classes. That, Ibelieve, should be tailored to individual programsand resources. However, I feel students should havesome understanding about the still and time beforethey can fully grasp the potential of interactivity.

The purpose of the following list is to highlight areasor directions that programs can follow in computergraphics. I chose to break this expansive disciplineinto four major areas; the still, time based works,interactive works, and computer graphics with othermedia. This same list could be viewed in many otherways, such as 2D, 3D, and time. The four suggestedareas are not mutually exclusive, as many crossoverin inextricable ways. Table 1 is a list of the fourareas and some topics that fall within each.

The Still Time Interactive

CrossComputerArt

2DImaging/Digital Paint

2Danimation

Web basedart

DigitalPhotography

3Dmodeling

3Danimation

InteractiveCDs

Printmaking &Computer

InstallationorSculpture

Digitalvideo

Interactiveinstallations

Fibers &Computer

Algorithmic

DigitalPerformance

VirtualEnvironments

Drawing/Painting &Computer

Table 1. Some suggested areas of focus incomputer graphics for the visual arts.

3.1. The StillThe still is a form of computer art that does notinclude time and can range from a print of a digitallymanipulated photographic image to an installation ora sculpture. The still can be wholly generated on thescreen, such as a digital painting or a 3D model, orcan include scanned or acquired imagery oralgorithms. Especially at issue with the still is“objecthood,” or the still as a physical material, suchas a print rather than an image on a screen.“Objecthood” is not a concern within the scope ofthis paper, however it is one that should be addressedin the course work.

The still is often thought of as a 2D or 3D image,however I have included installation and sculpturalworks because they deal with a single instance oftime. An understanding of 4D is so crucial to someareas in computer art that I used it as an organizingfactor that surmounts the difference between imageryand sculpture. A digital installation could be timebased, especially if it includes digital video, but thesculpture I refer to in the still category is one thatdeals with a single instance in time. This includesanything from the latest resin basedstereolithographic sculpture to still images that havebeen realized in some sculptural form such as digitalimages wrapped around a cube.

3.2. TimeIt is one thing to understand how the composition ofa single image works formally and expressively, butit is quite another issue to understand how thataesthetic changes over time. What is important forexpression when the viewer cannot contemplate eachframe? How do ideas change as the still imagemoves? How does time factor into the overallsuccess of the art? These questions and more areones that time based course work must address.

Courses in this area include 2D and 3D animation,digital video, and performance using computergraphics technologies. Students who focus in thisarea should take support courses in theater, film, anddrama.

3.3. Interactive WorksSince The Dream Curriculum the notion ofinteractivity has grown profoundly more complex,especially in light of developments such as theWWW and multi-user installations. These coursesinclude web based art, interactive CDs, interactiveinstallations, and VEs.

Students who work with interactivity use both timeand imagery as well as human response. This coursework should include an understanding of humancomputer interface design as well as concepts suchas the aesthetic experience for viewers as they unfoldlayers of the work.

3.4. Computer Graphics and other MediaThe ways in which computer art crosses into otherart areas is abundant and numbers far beyond thefour that I have listed. Computer graphics as anextension of other media is helping artists approachtheir studio area in a new way by enlarging the scopeof artistic possibilities. For example a painter canscan a slide of her painting, manipulate it, reproduceit as an Iris print, then transfer it to wet plaster tocreate a sculpture with a fresco aesthetic. Thisprocess surpasses the act of painting and perhapsdefies a label, but it is a unique process madepossible by computer graphics. It is this type ofthing that would fall within the Computer graphicsand other media focus area.

In addition to computer art, students in this areashould have an equally strong focus in at least oneother studio area.

4. Conclusion

In a time when our culture is becoming much morevisually literate, those who create visual media mustpay more attention to clear, meaningful, andsophisticated messages in their work. Visual mediafor the sake of effect stops at the surface of the work,and the audience is lost after the initial “wow.” Thisis much like movies that sport fascinating specialeffects but fall short of a meaningful story line. Onthe other hand, movies that weave monumentalvisual effects into purposeful narratives are theworks of art that will be significant in years to come.This means that the curriculum that educates some ofthe artists who will make these films must addresstechnical proficiency along with creativity andmeaningful expression.

To this end I have highlighted a few concerns in corecurriculum, listed some computer art focus areas,and suggested key issues to address in the generalapproach to designing curriculum for computergraphics in the visual arts. These are onlysuggestions and individual programs and goalsshould be the ultimate guiding force.

My hope with this document is to continue thedialogue about curriculum so computer graphics

educators in the visual arts can address it in ameaningful and vital way.

References

[1] H. Gardner. Frames of Mind: The Theoryof Multiple Intelligences. Basic Books, Inc.,New York, 1983.

[2] J. Harris. Art Education and Cyber-Idealogy: Beyond Individualism andTechnological Determinism. In Art Journal,pages 39-45, Fall, 1997.

[3] T. A. Jackson. Ontological Shifts in StudioArt Education: Emergent pedagogicalModels. In Art Journal, pages 69-73, Spring1999.

[4] P. Klein. Structuring Digital Media in aFine Arts Institution. Paper presented to theannual meeting of the College ArtAssociation, February, 1999.

[5] M. Kurlansky. Educating Designers in theDigital Age. In Digital Creativity. Lisse,

the Netherlands, Swets and Zeitlinger, pages218-222, Vol. 9, No. 4, 1998.

[6] M. Mcluhan. Myth and Mass Media. In M.Moos Media Research: Technology, Art,Communication. OPO, Amsterdam, pages5-15, 1997.

[7] B. Mones-Hattal, K. O'Connell and D.Sokolove. Guidelines for Curricula inComputer Graphics in the Visual Arts. InComputer Graphics, pages 78-76, June1990.

[8] L. Tolstoy. Extracts from What is art? In E.Simmons, Leo Tolstoy, VII. Vintage Books,New York, page 239, 1960. (Reprintedfrom What is art?, (Aylmer Maude Trans.).by L. Tolstoy, Oxford University Press,Oxford, 1898)

An Interdisciplinary Approach to Teaching Computer Animation toArtists and Computer Scientists

David S. Ebert and Dan BaileyUniversity of Maryland, Baltimore County

{ebert, bailey}@umbc.edu

Abstract

Animation has always required a closecollaboration between artists and scientists, poetsand engineers. Current trends in computeranimation have made successful and effectiveteamwork a necessity. To address these issues, wehave developed an interdisciplinary computeranimation course for artists and scientists thatfocuses on contemporary issues in computeranimation. This course emphasizes collaborativeteams for practical experience, cross-mixing ofstudent expertise, and group-based education: thestudents learn from each other, as well as theinstructors. Student teams produce a professionalanimation that extends the capabilities of acommercial animation package to gain experiencein and exposure to the state-of-the-art research incomputer animation and rendering, the completeanimation process, and the artistic and aestheticsof computer animation. We describe our approachto teaching this course, the structure of the course,the results, and lessons learned from ourexperience with the first offering of this course(Spring 1999).Keywords: interdisciplinary, computer animation,education, collaborative education.

1. Introduction

Computer animation has always required a closecollaboration between artists and computerscientists. We have developed an interdisciplinarycomputer animation course that focuses oncontemporary issues in computer animation andrequires the skills of animators and programmersworking in teams. The goals of this course are thefollowing:• Increase the technical graphics and animation

knowledge of computer science students.• Increase the animation skills and knowledge

of advanced computer animation techniquesof art students.

• Introduce art students to the technical aspectsof rendering and animation, and expose themto research issues in computer animation.

• Introduce computer science students totraditional and computer animationtechniques.

• Introduce art students to the creative potentialof writing procedural shaders, models, andanimation expressions.

• Provide practical animation productionexperience, using and extending commercialanimation software.

• Provide a collaborative learning environmentwhere students will learn from each other, aswell as the course instructors.

A key aspect of this course is that students gainexperience in participating in interdisciplinaryteams. Teams of 4 - 5 visual arts and computerscience students work together to produceanimations that utilize each member's skills andinterests, in a manner similar to commercialanimation environments. The computer animationindustry requires employees to work in teams onlarge projects. Traditional educationalenvironments do not teach skills to make studentssuccessful in such environments. Computeranimation environments pose a further challenge:teams are composed of members from quitedisjoint backgrounds. We have structured thiscourse to help students learn how to communicate,work, and even thrive in this environment.

2. Background and Motivation

Animation's history, from its origins in the 1880sto contemporary time, is a continuous line oftechnological inventions that have allowedanimators the ability to achieve higher qualityeffects with greater ease: Bray/Hurd patented cels,Fleischer Brothers invented and patentedrotoscoping, Disney developed the multi-planecamera, and the list goes on [1]. At the heart ofeach of these developments has been thesuccessful synthesis of artistic and scientifictalents. At times, these skills have come togetherin the unique individual. Normally, however, theyhave been the result of creative collaborations,especially in today's highly technical computeranimation arena.

Equally as important, large scale animationproduction has always required large teams ofvariously talented individuals. The WarnerBrothers animators of the 1930s through 1950sarguably produced some of the most successfulcartoons of this century and historians note thateffective collaboration between its directors,animators, writers, technicians, artists, andmusicians was one of the prime reasons for itssuccess [1].

Both of these issues point to the fact that asuccessful and contemporary animationcurriculum should not only be interdisciplinary,but also encourage students to develop effectiveteam skills.

Schools, universities and institutions are usuallydivided into departments to better serve studentsneeds. Crossing the boundaries between areas anddepartments has always been difficult. Aparticular school may not be reluctant to developinterdisciplinary courses, but usually mechanismsand incentives to do so are not in place. Thus,many schools are slow to address the industrytrend to teach and encourage effective teamworkand collaboration between animators and computerscientists. Currently many animation, specialeffects, and computer graphics houses are creatingtheir own in-house workshops and programs foraddressing these issues.

3. Our Pedagogical Approach

Our pedagogical approach to this course has twokey themes: interdisciplinary work andcollaborative education. This is true even in theinstruction and design of the course, which is teamtaught by a visual arts faculty member and acomputer science faculty member. Most of ourlectures are designed to have sections that bothfaculty members present, highlighting thetechnical computer graphics aspects and theart/animation aspects of the material. In everyaspect of the course, we encourage students tocollaborate and help each other. Initialassignments, described below, are designed tohave segments that are easy and segments that aredifficult for students from each background, thusencouraging students to begin interacting withtheir counterparts.

As mentioned above, we are presenting material inthe course that extends the backgrounds of bothstudents, while making them familiar with (and

even extending) the state-of-the-art in computeranimation. Computer science students lead thediscussion of computer animation research papers,helping the art students understand the newmaterial. Conversely, art students present profilesand critiques of computer animators and animationtechniques that expand the computer sciencestudents' appreciation of computer animation asart.

4. Structure and Implementation

We have structured this course to take advantageof a classroom equipped with SGI workstations forinteractive instruction and demonstrations, and theopen architecture and procedural flexibility of theMaya software package from Alias/Wavefront1.The course is 15 weeks in length and the studentsstart working in teams (4-5 students) during thesecond week of the class. The students maypropose the composition of their team, with therestriction that each team must consist of at leasttwo art students and two computer sciencestudents. The students have an icebreaker teamproject to perform in Weeks 3-5 of the class:create a 10-second animation to be compositedwith the blue-screen filmed sequence of a studentperforming for 10 seconds. After this initialicebreaker project, we allow the teams to berevised to accommodate any poor group dynamicsthat were discovered. We also have some initialassignments that expose artists to working withvectors, angles, and simple illumination, exposecomputer scientists to key-frame animation, andboth to the procedural, extensible aspects of theMaya modeling, animation, and renderingpackage.

To expose students to the power of proceduralshading techniques [2] and the challenge ofphotorealistic image generation, the students'second major project is to generate a photorealisticimage/animation with the following specification:it must contain a specific type of light and aspecific object element each chosen randomlyfrom a hat. Some example lighting types used inSpring 1999, included sunset light, light from alava lamp, light from a neon sign, and light from acandle flame. Some example object elementsincluded a cup of cappuccino with froth, icy orsnowy patches on a sidewalk, and fur. Studentswere given three weeks to complete thisassignment.

1 This software was made available through agenerous grant from Alias/Wavefront, Inc.

The main component of the course is a 12-weekteam project to produce an interesting,professional animation that includes extending theMaya package in a new way. Some proposedextensions include the following:• Particle dynamics for cloud formation• Volume rendered implicit primitives for

accurate cloud rendering• Volumetric light sources to model fire• Volumetric simulations of fire, taking into

account simple physical simulations• Texturing of implicit surfaces with

deformations• Extending Maya's implicit animation

capabilities• Hydrologic simulations of ocean waves• Underwater simulation of light rays and

dispersion• Simulations of rich, organic, evolving

environments• Flocking simulations for crowd animation

Regardless of the technical and graphic effects thatare achieved, a successful animation is only asgood as its story, premise, or content. Studentsmust also consider the subject matter of theproject. Suggested formats and structures for theanimation include the following:

• Narrative (comedy, drama, adventure. etc.)• Character driven• Animating directly to sound or music• Environmental or architectural (a precursor to

a VR or interactive space)• Poetic or text driven piece (using a quotation,

journal entry, voice over, or written text)• Commentary (political, technological, social,

theoretical)• Adaptation of existing works (homage,

contrast, contradiction, satire)

Early in the semester each team is required to"pitch" their animation to the instructors and theentire class in a professional presentation withstoryboards, charts, and slides. This project alsorequires a presentation of progress to the classafter five weeks of work and a final presentation ofresults to the class. These are graded both on thepresentation, and the artistic and technical merit ofthe work.

5. Initial Results

The initial offering of the course was verysuccessful and educational for the students and

instructors. The students were very enthused aboutthe class and gained valuable experience incomputer animation, as well as working in teams.Most of the teams worked successfully togetherand, to our surprise, each team decided to nameitself. The team dynamics varied widely amongthe 5 teams and also changed greatly from thebeginning of the semester to the end. One of theteams that worked the best during the semesterstarted out with very poor group dynamicsbetween the artists and computer scientists. By theend of the semester, they learned how tocommunicate effectively, to appreciate eachother's skills, and work as a team. Only one team(20%) suffered significantly from group dynamics,including one team member dropping the class.Spending more class time on effective teamtechniques and team building may help eliminatethese problems.

5.1. Blue-screen ProjectOur icebreaker team project required each team tolearn how to composite live-action video (10seconds of a student using exercise equipment)with computer generated effects for a 10 secondanimation. They were required to generate bothCG foreground and background elements. Anexample still of the blue-screen video is shown inFigure 1. Two example stills from this project areshown in Figures 2 and 3.

After viewing and critiquing the students' initialanimations, we gave them an additional week toimprove them. This produced a better product, butalso affected the amount of time available for therest of the semester work. Students spent 5 weekscompleting this project. After viewing the finalanimations, the importance of teaching anti-aliasing techniques became apparent.

5.2 Artistic Shader AssignmentAs mentioned above, the goal of this assignmentwas to create a photorealistic image or animationthat contained a specific type of lighting and aspecific element. After each team chose a type oflight and a scene element from a "hat", thefollowing became the scene specification for eachteam:

Team1 Kiwi and peach in a bowl Candle lightTeam2 Fur Mirrored disco ball

and strobe lightTeam3 Hot tub with steam and

bubblesNeon light with neontube showing

Team4 Ice patches on a sidewalk Outdoor sunset orsunrise lighting

Team5 Cappuccino with frothand a cinnamon swirl

Lava lamp light

Figures 4 and 5 show two example stills from thisproject. Figure 4 contains an image showingsnowy patches on a sidewalk at sunrise and Figure

5 contains an image of a cup of cappuccinoilluminated by a lava lamp.

5.3 Semester Animation ProjectThe semester animation projects contained a widevariety of technical and artistic styles. Below aresummaries of the projects for each team:

• The Autonomous Chicken Farm - This teamexplored the practical application ofbehavioral animation by developingprocedurally animated "Creatures" which arecontrolled by a customized user interface inMaya. The example "Creatures" chosen werechickens with controllable personalityattributes of hunger, beauty, and aggression(based on the 7 deadly sins). The final productof this project was an extension of Maya forcreating autonomous "smart" creatures and

Figure 3: A still from the blue-screenanimation by CSmART Allstars (AvaCollins, Alex Eller, Jason Lubawski,Marlin Rowley, Christian Valiente).

Figure 2: A still from the blue-screenanimation by the Screaming Nixons (EunBaek, Jon Feibelman, Costas Kleopa,Vlad Korolev, Steve Matuszek). Figure 4: Snowy sidewalk at sunrise by the

Screaming Nixons (Eun Baek, JonFeibelman, Costas Kleopa, Vlad Korolev,Steve Matuszek).

Figure 5: "Cappuccino by lava lamp light,"by CsmArt Allstars (Ava Collins, Alex Eller,Jason Lubawski, Marlin Rowley, ChristianValiente).

controlling their movement. This generatedinput for the inverse-kinematic controls andexpressions that the animators used forcontrolling the motion of the creatures. Ashort demonstration of the Maya extensionwas also produced.

• Gelegant Jammin' - Animated short in whichcharacter movements (blendshapes, joints, andlattices) of a drop of gel were driven by soundfile information (wav file). A graphical userinterface was created to control which dancemovements were driven by particular sounddata.

• Infinity - This animation was a symbolicprojection of life and astral time representedthrough the juxtaposition of an angelcontemplating notions of time, history andevolution, with an icon of modern technology.Plugins for the rendering and modeling ofvolumetric clouds were developed to set thescene for this animation.

• Martyr's Playtime Theatre Presents: Joand'Arc - The story of Joan of Arc waspresented using models resembling thechildren's toys, the Fisher Pricetm People.Procedural effects were created for fire,melting, and crowd animation.

• A Buck Ninety-Nine and Someone Else's Wall- This project explored the ambiguity ofenvironment by moving between caveformation space and the urban realm of thesubway station. The main character was aspelunker who wears a headlight to illuminatethe scenes. Initially, he seems to be locatedalone in a cave somewhere, but as he exploresthe environment he finds himself to really bein the tunnels of the subway system. In orderto create the cave space, a MEL (MayaEmbedded Language) script was created forcreating turbulent cave walls with varyingfractal parameters.

Some example resulting images from theseprojects can be seen in Figures 6, 7 and 8. Mostteams completed a significant portion of theirproject, but didn't complete it. We worked with theteams to complete their animations after thesemester so that they had produced a finishedanimation from the class that they could besatisfied with and include in a demo reel. Teamshad difficulties both in terms of the technicalprogramming aspects, as well as animation andproduction.

6. Conclusions

We have developed a successful interdisciplinarycourse to teach computer animation to computerscientists and artists based on interdisciplinarycollaborative work. This approach for education isvery powerful and rewarding; however, it doesrequire a significant amount of effort in teaching

Figure 6: A still from Infinity by theCSmART Allstars (Ava Collins, AlexEller, Jason Lubawski, Marlin Rowley,Christian Valiente) showing an angel ina volumetric procedural clouds.

Figure 7: A still from Martyr's PlaytimeTheatre Presents: Joan D'Arc by theScreaming Nixons (Eun Baek, JonFeibelman, Costas Kleopa, VladKorolev, Steve Matuszek).

not only computer animation, but successful teamwork techniques and group dynamics.

Our first offering of this course was verysuccessful and we have gained valuableexperience to help us improve the course in thefuture. For the next offering of the course, we willdecrease the amount of time students have tocomplete their first icebreaker assignment so thatthey can concentrate their efforts on their semesterprojects. We also will allow for more individualcredit in the team projects. Most of a student'sgrade in Spring 1999 was based on the grade theirteam received. We made a few exception inextreme cases. We plan on still having most of thework in the class be done with collaborative teams,but will be adjusting the grading scheme to alsoreflect each individual's contribution. Thehomepage for the Spring 1999 edition of this classcan be found athttp://research.umbc.edu/~bailey/courses/CS_ART_Anim/

From our experience, we believe that aninterdisciplinary computer animation courseprovides both types of students with an increasedappreciation of, and an improved ability tocommunicate with, the other community. Studentsalso gain valuable experience in working incollaborative teams and an increased sense of thehistory and state-of-the-art of computer animation.

We would like to thank the student teams whohelped us develop and improve this course:

• The A-Team (Tracy Corder, Will Gee, MikeKeesey, Joe Romano)

• CSmART Allstars (Ava Collins, Alex Eller,Jason Lubawski, Marlin Rowley. ChristianValiente)

• Beasts (Kim Harrington, Mike Madison, ChrisMorris, Brian Resurreccion, Shawn Wood)

• Screaming Nixons (Eun Baek, Jon Feibelman,Costas Kleopa, Vlad Korolev, SteveMatuszek)

• Wookie Pimps (Michelle Hunt, Steve Jacobs,Chris Slingluff, Joy Saunders)

References

[1] Maltin, L., Of Mice and Men, McGraw-Hill,1980.

[2] Ebert, D., Musgrave, F., Peachey, D., Perlin,P., Worley, S., Texturing and Modeling: AProcedural Approach, second edition, APProfessional, 1998.

Combining Art and Mathematics in Computer Graphics Education

Mark OllilaUniversity of Gävle

Gävle, Swedenmolly@hig.se

Abstract

This paper presents an innovative education programat the University of Gävle, Sweden. The program,Creative Programming, combines traditional methodsof teaching computer graphics, with other cross-disciplinary areas such as art, cognition, and film. Theuniqueness of the program is represented by thestudents that are accepted and the teachingenvironment they are put in. A judicial process is usedto screen applicants, with and without portfolios to theprogram. This is followed by an examination ofintellectual capacity, practical examination, and finallyan interview. Together, these are used as guidelines todetermine if the student is a suitable candidate for theprogram. The final group selected has varyingbackgrounds, ranging from the professional artist tothe traditional computer scientist. They are thenimmersed in a lab environment for the whole durationof the education on the campus. The remainder of theeducation involves a period in industry. Thecombination of selective students, a lab environmenthas produced very satisfactory result, and can be usedas basis for further developing the education.Keywords: Computer Graphics Education, Art,Industry

Introduction

Universities are under pressure to develop new,dynamic education curricula. The area of digital media,for instance, is becoming increasingly important in thisworld. There is a shortage of people with the requirededucational background who can design systems in thisnew medium, in research terms, as well as in visualdevelopment and implementation. The interfaces tocomputers are designed in a much more visual andintuitive manner today, so the skills involved increating software applications have expanded fromtraditional programming to include creative skills, suchas graphic design, writing interactive narratives andcreating story boards. Research also plays a veryimportant role in the design of systems andimplementation of new techniques yet to be utilized incurrent technologies. Universities with traditionaleducation of graduates in computer science must nowexamine the need for these new multidisciplinary

skills. One common solution is the creation of cross-disciplinary degrees such as Computer Science/MediaStudies. If we examine the skill set required indeveloping these new media applications we find thatthey are essentially different depending on the desiredend result. Several applications will require aprogramming/logical approach as a solution, whereasothers will require a creative, visual solution to theproblem. If we examine the various types of graduatesfrom a typical University, we will find different skillsets appropriate to this new medium. An arts graduatewill have developed skills in written communication,an artist will have an appreciation of aesthetics,psychology graduates appreciate the difficulties ofhuman-computer interaction, and computer sciencegraduates will have an understanding of the technicalissues required and how current technologies might notbe adequate for the problem at hand.

The Creative Programming program was created in1995 with the goal of combining the previouslymentioned skill sets together in an innovative way.The initial program is an eighteen month programwhich can be used in conjunction with computerscience credits (or similar) to obtain a Bachelorsdegree. The program looks at both the technical andaesthetic issues involved in the digital media of today.Education is in the form of lectures and practicalsessions with skilled instructors from industry oracademia. This makes the students up to date with whatis happening in industry, as well as understanding whatthe latest developments in research are. The programhelps students to combine existing skills with thetechnological skills needed in this area, enhancing theirknowledge, both technically and analytically.

Screening Process

The process of screening applicants is very complex.The work is examined by a jury of experts in the areaof digital media, both from academia and fromindustry. A previous Creative Programming student isalso included in the jury to provide a studentsperspective. The jury reduces all the initial submissionsto thirty possible candidates that are then invited for aninterview at the University. The criteria used forselection include the portfolio, academic qualifications,

and industry experience. When judging the portfolio,issues of creativity, originality and quality are mostimportant. It is often the case that the portfoliosexamined are varied in nature, presenting very differentkinds of work, in either creativity and originality, butwith not necessarily high technical art skill.

The interview generally involves an examination of theapplicants intellectual ability, creative ability, skill athandling a computer, and evidence of a creative mindas well as of sociability. One element of the interviewstage is the use of Raven’s Advanced ProgressiveMatrices (APM) [1] to indicate the analytical andlogical ability of the applicant. The APM is designed toassess a person's intellectual and reasoning ability. It isone of the most well examined tests of generalintelligence, it is nonverbal, and has very goodpsychometric properties, normated against lots ofpopulations with different cultural and educationalbackgrounds. The advanced version of the test (APM)used here is designed for people of above averageintellectual capability. The test-retest reliability is overto 0.93 for age 30 and under. The test is used forevaluating the logical intellectual capacity of thosevery artistic applicants that are lacking traditionalcomputer science. The social and emotionalcompetence is also judged in the interview, to makesure that an open and dynamic group process ispromoted by the students accepted. The creation of thisgroup dynamics is also the rationale for selecting agood mix of gender, age and artistic and computationalprofiles into the group. With this the most crucialconditions for merging the artist with the programmerare met with.

The program is unique in its approach to education andselection of students. It aims to give an opportunity tohighly creative and intelligent people to experimentand enhance themselves in the world of digital mediaby combining their skills from other areas, such as artor programming.

Curriculum

After the selection process, fifteen students are invitedto participate in the Creative Programming program.The first twelve months of the program the students arein a lab environment at the University. The final sixmonths the students are in industry. The program istaught in competition to the Computer Scienceprogram and the Computer Engineering Program at theUniversity. The structure of the curriculum is taught infour terms over two semesters, with the final industryproject in the third semester. The first semesterconcentrates on aesthetics of the image, new media

technologies and human computer interaction. Thesubjects taught in that semester include -

Multimedia Techniques – introduction to the theoreticalconcepts of multimedia. The practical sessions involvedeveloping multimedia applications, which are bothtechnically and aesthetically challenging. This is taughtby Computer Science staff.

Aesthetics – introduction to the important concepts inaesthetics. This looks at both traditional and currenttheories in aesthetics. This is taught by an expert in artfrom a department at the University. There is also someinteraction from industry art experts.

Multimedia Management – this subject examines howto create complex multimedia applications. This iscombined with dealing with real clients and deadlines.The clients are usually from industry or the localcommunity. The lectures involve seminars about topicssuch as contract writing, copyright issues, managementof a team and client handling. These seminars are givenby experts from the digital media industry.

Program Design and Cognition – examines the humancomputer interaction (HCI) issues when designingsoftware. There is a strong cognitive psychologybackground presented here. The course is taught bylecturers in the Computer Science department.

The second semester of the course concentrates oncomputer graphics, animation, film aesthetics andfuture research. The subjects include -

Computer Graphics, Modelling and Animation – thiscourse is actually two courses in one. It examines thetheoretical aspects of computer graphics in detail. Eventhough the majority of students do not have thetraditional linear algebra background, the course stilldelivers complex mathematical concepts. The contentsinclude, but are not limited to, 2D and 3Dtransformations, lighting, realism, renderingtechniques, shading models, colour models, hiddensurface algorithms, 2D algorithms, 3D algorithms,radiosity, surface representations, facial animation,facial modelling, volume rendering, virtual reality,future research. The practical sessions are taught bycomputer animators from the industry. This will ofteninclude someone to handle the creative direction of ananimation as well as someone to deal with the technicalissues. This is to reinforce what the students have beentaught in lectures. The lectures and programmingoriented practicals are taught by Computer Sciencestaff. The modelling and animation, creative directionaspects are taught by appropriate experts in theindustry.

Film Aesthetics – this is a course taught by the Mediadepartment at the University. It examines concepts ofwhat is film, the types of film, criticism in film. It istaught by various media professors and improves thestudents’ ability to analyse work. The course also helpsthe students in designing storyboards and narratives.

Industry Seminars – this course is the last before thestudents have their industry practice. Approximatelyten industry experts are invited to give seminars onhow digital media is affecting them. These lecturershave included people from film, government, post-production, computer gaming, research, smallcompanies, large companies and so forth. The maingoal is to avoid presenting a narrow view of the digitalmedia industry. The students also have a free project oftheir choice to deliver.

In all the above courses, the students have access totheir own lab of equipment. The lab usually consists offifteen personal computers, six SGI computers, onescanning machine, two printers, a video editingmachine, TV’s, video equipment, virtual realityequipment and so forth. This equipment is accessible24 hours of the day to both staff and students in theprogram. The benefit of the lab environment is that itacts as a place where the students can learn from eachother and to develop.

Practice versus Theory

The Creative Programming program examines manyissues of what is important in content in today's digitalmedia. This involves highly theoretical lectures inareas such as HCI, modelling and animation, andcomputer graphics. The general idea is that the theorydoes not change remarkably, but the tools do.Therefore, it is important to stress the theory more thanthe tools. This has become obvious in several instancessuch as the 3D animation industry, where softwareupdates and competition is fierce. The important thingthat is stressed in the education is that the tools areirrelevant. With good support of industry we are able toachieve this high standard of education, withoutcompromising the principles of quality with quantity.

HCI is a very important part of computer graphics, andhence the program involves a heavy cognitive theorysection. As we see in Figure 1, simple design tasks aregiven to the students to help build up an understandingof simple, easy to use interfaces. When educating bothprogrammers and artists in computer graphics, apragmatic approach is required. We stress the theoryaspects of computer graphics, modelling and animationto both students, as these principles generally will

remain the same. The practical aspects of the courseare performed by industry, and as of such, we do notstress the tools that we use, only how the theory affectsthe tools. Looking at Figures 2 and 3, we can see someexperiments with lighting, reflection maps, andradiosity concepts using the tools available in the labs.This is stressed in the lectures, and the students areencouraged to experiment with the theoretical ideaspresented.

Figure 1. The theoretical concepts of Design andCognition are used to help students to designsimple and language independent interfaces. Inthis particular example, a new photocopierinterface is designed.

Group projects are often encouraged to educate thestudents about collaboration. There is also plenty oftime for students to perform individual projects. Duringthe Computer Graphics, Modelling and Animationcourse, the students are strongly encouraged to submittheir work to various international competitions. Theseinclude the Rhythm and Hues student competition, theAlias student competition for instance.

Figure 2: A still from an animation to show thefuture dental clinic. This is a project with realdeadlines and real clients.

Figure 3: A still from a two minute animationpresenting the future dental clinic. This wasproduced after a highly theoretical lecture onradiosity.

Industry Involvement

The final stages of the students’ education is spent inindustry, where the student specializes in some area.Generally, this has included programming in thegaming industry, 3D industry (animation andmodelling), and also the web industry. The process ofmatching the student to industry is often a complextask. This usually involves various visits by the studentto different companies that they like. This iscomplemented by visits by companies to the lab wherethey are able to see the student at work. When asuccessful company has been found, the student thencan begin their practice there. Some conditions arestressed by the University about the practice session.These include that the student be guaranteed acomputer and desk, that they are working on someproject, and that they are given one day a week toprepare a report and presentation. At the end ofpractice (twenty weeks), the student then presents amajor presentation, which is either research related orpractice related. To ensure that both the student and theclient are satisfied constant communication by thesubject controller to both parties is maintained. It is inthis process that feedback about both client and studentis received. This is important as it allows us to reviewthe selection process for both the student and thecompany. At the time of writing, we are about to sendout our fourth group of students into industry. Thisprocess and relationship with industry was discussed in[2]. In Figure 4 we can see the result of one student’swork whilst in industry. The student worked in thespecial effects industry and produced over fiveproductions that were shown on television. Figure 5and 6 demonstrate how the movement from traditionalart to computer generated art can produce very goodresults. The art work in Figure 5 was in part of coursework. The model in Figure 6 was individual free timework that was applied after various theoretical lectureson facial modelling. Figure 7 is a very complex model

that has been submitted to the Rhythm and Huesinternational student competition.

Figure 4: A still from the industry project where astudent worked on a commercial that waspresented on television.

Figure 5: Traditional artwork.

Figure 6: Facial modelling by same student as infigure 5.

Figure 7: A rendered model with advancedtextures.

Conclusion

In this paper a new digital media program is described,one that attempts to merge the computer programmerand artist together. The program aims to provide theopportunity for the programmer that has artistic skillsto combine them in an education, and for the artist thathas never approached a computer, to learn and improvethemselves. As APM is a very good predictor ofacademic success, it can also be used to rise theconfidence of this group of applicants in that they willmanage the theoretical and technical courses in theprogram. An examination into the methods used incombining theory of Computer Graphics with practiceis also explored. A lab environment is used, where thedifferent cultures of people, experiment with newtechniques, and use each other as reference ofinformation, both technical and aesthetic wise. Thismeans, those that are familiar with computerprogramming, can educate the artists in some aspects,and in return receive aesthetic criticism and tutorage.The education as such, concentrates on theoreticalaspects of multimedia, design and cognition, computergraphics, modeling and animation, aesthetics, andmedia issues. The practical side generally involvesreinforcing the theory, but also acts as an avenue forthe students to enhance their creativity. Industryinvolvement is also crucial, and hence the studentsoften find their practical work has a real end purpose.

Acknowledgements

Students Torsten Edwinson (figure 1), Mattias Malmer(figures 2,3 and 7), Hanne Drossel (figures 2,3),Mikael Håkansson (figure 4), Ulf Lundgren (figure 5

and 6). Eva Carling for her comments regarding theRavens test.References

[1] Raven, J.C. (1993). Advanced ProgressiveMatrices, Sets I & II. 1994 edition, with newnorms for UK and USA adults and othernational groups. Oxford Psychologists Press.

[2] M. Ollila, J. Kempff, and J. Ljungman.Where Industry and Academia meet : aninternational perspective. SIGGRAPH 98Panel Presentation, Orlando Florida.

Computers in Art an d Design Education — Past, Present and Future.

Tony LongsonCalifornia State UniversityLos Angeles, California.

tony@design.calstatela.edu

Abstract

A backward look at the use of computers in Art andDesign Education is used to speculate about the futurein an attempt to anticipate and determine events. Theauthor sites practical examples, and raises morequestions than he knows how to answer.

Keywords: Art and Design education, technology,software-packages, Object-Oriented, creativity,problem-solving, browser, collaboration, tailoredenvironment.

1. Introduction

Design consists largely of problem solving. In a 1972interview [1] Charles Eames answers the question“What are the boundaries of design?” with anotherquestion - “What are the boundaries of problems?”

Art too involves problem solving. What distinguishes itfrom design and promotes it to one of Mankind’shighest endeavors is that an Artist also has to invent theproblem to be solved.

Computers are problem solving tools and, if anyjustification were needed, this alone is sufficient forusing them in the visual arts.

In order to solve a problem the computer needs to beprogrammed. This provides a creative potential for theartist or designer - visual problems need to beexpressed in a structured way through a language withits control structures and variables, algorithms need tobe invented. The process requires one to think of thecreative problem in new ways and in doing soinnovative creative solutions will inevitably show up.

2. PAST (1960’s - 1986)

I started to use computers in my art in 1973 and wasannoyed to find that they could make better drawingsin seconds than I could make in weeks. There was aprecious quality to my drawings that the computer hadno regard for. Taking the skill out of my work mademe look more closely at the content. My secondthought was a feeling of excitement that I could define

rules that would condition what the image would be.My arts heritage was the English Constructivistmovement, the Systems Group and theConstructionists. And my third thought (all this withinan hour or so) was that I had to teach my students howto use computers in their work.

The tools at our disposal were limited: BASIC orFORTRAN with simple device control such as movetoand lineto, output was a small green CRT or, when theink didn’t run dry, a pen plotter.

Despite, or perhaps because of, the limitations, thework that people produced was extraordinary both inits originality and its integrity. Examples of this workwere shown in the Siggraph ‘86 Art Show curated byPatric Prince[2].

I was visiting lecturer to London’s Slade SchoolPostgraduate Experimental Area lead by MalcolmHughes. A handful of students working with aneviscerated Data General mini-computer includedChris Briscoe, who pioneered commercial computeranimation in England, and Paul Brown who has been amajor influence in education. Brown’s early work canbe seen at his web site.

The tools settled out into three categories: 2D graphicspackages, 3D graphics packages and ‘Paint Programs’.The first two were really the domain of scientists andengineers as they worked in conjunction with aprogramming language. The Paint Programs weresomething of an anomaly. Developed by scientists whopresumably had some idea that they would be useful itwas up to artists to demonstrate the fact.

While people were asking whether computers could beused creatively (Yes), artists like David Em (JetPropulsion Laboratory) and Darcy Gerbarg (New YorkInstitute of Technology) were quietly getting on withit. The Paint Program at JPL was written by Jim Blinnwho used it to make texture maps for the moons andplanets in his simulation of the solar system. As Irecall, he usually got his brother (an artist) to do thework.

The next question up for discussion was “should artistsand designers know how to program?” I said yes, andwas derided for it! In fact I had already established twosuccessful educational initiatives in ComputerGraphics for Artists and Designers based on thatpremise. One was at West Coast University in LosAngeles (now defunct), and the other - where I stillteach - is in the Art Department at California StateUniversity, Los Angeles. I think they were the first oftheir kind. People were lining up to get in [3].

Frankly it wasn’t too hard to make that decision. I hadtaught myself to program and enjoyed the uniquefeeling of control over creative questions that itoffered; I could see excellent work from other artists;and... there wasn’t anything else to teach! Softwarepackages didn’t exist and the use of computers inDesign was just beginning. In 1984 the Macintoshsuggested a potential for design but didn’t deliver until1986.

3. PRESENT (1986-now)

It was the invention of the Laser Printer and its pagedescription language, Postscript, that delivered thepromise of computers in Design -_Output.

Desktop Publishing became a practicality. Commercialsuccess lead to cheaper computers. Cheaper memorylead to more colors (one of the rare examples in lifewhere you get more out than you put in,) more colorscreated a need for paint programs, 2D and then 3Dmodeling and animation packages became common,and HyperCard established a practical basis forinteractive design. Specialized software packagesproliferated - and that is what people began to teach.

Why not teach software packages? They offer a well-defined body of information, usually come withdocumentation and tutorials, and are easy to justify toadministration. Furthermore there are hundreds of jobadverts which say “must know PageMaker”, andhundreds of students who want to get that job. As aconsequence you’ll see academic courses inUniversities across the world with titles like“Advanced Photoshop”. It sounds reasonable but it is amistake.

Academic courses in Art and Design are aboutcreativity and problem solving, the tools are secondary.Also the tools are out of date in a few months, so that astudent who know menus and commands and keyboardshortcuts for one program will be at a loss when thenew version comes out.An alternative is to teach the principles behind thesoftware and to set this in a creative context where

practical projects encourage an exploration andunderstanding of the concepts. The principles behindthe software will survive. Teaching a student theprinciples of computer graphics means that she will beable to transfer her knowledge to whateverenvironment of software and hardware she findsherself in.

Computer Graphics software seems to separate out intocategories: drawing programs such as Illustrator orFreehand characterized by a 2D coordinate system withtransformations operating on object-oriented lines andshapes; raster based graphics such as Photoshopcharacterized by pixel-level manipulation, layers, andimage-processing techniques; 3D modeling andanimation programs such as CINEMA4D or Mayacharacterized by 3D coordinate systems, 3D objectswith transformations applied to them, rendering, andthe notion of change over time; and Interactive Designwhich may include any of the other categories but ischaracterized by non-linear presentation of ideas andinformation in a context which is easy to negotiate.

It is not surprising that these software packages hadtheir origins in the 2D, 3D and paint programs of theearly days. Even Interactive Design, proposed by TedNelson in the 1960’s, had most of its design principlesraised and codified before the Internet made it a reality.

I try to tailor projects to encourage an exploration andunderstanding of the principles behind the software. Imake the projects “self-starting”. Creativity is a cycleof activity and review. Often students find it difficult tostart the process so I invent projects which will havethem producing objects and images without mucheffort. Usually the projects are rule-based. When theyhave an image in front of them that has not requiredtoo much effort on their part students are usually moreopen to critically evaluating it and making changes,and so they quickly become involved in the creativeprocess. Once the image or object has a life of its ownthe students can respond to it creatively.

I also try to invent projects which will allow thestudents to juxtapose their work with someone (oreveryone) else in the class.

You might agree that the idea of teaching principles(not packages) is good but argue that at some point thestudent needs to be shown details of the softwarepackage such as which menu item to select or windowto open. I would counter that if the student can’t figureit out for herself then there is something wrong withthe software package. As consumers we wouldn’tdream of buying a domestic appliance such as awashing machine or a car whose operation was not

self-evident. Software should be the same. I usuallytake a naive approach to teaching software explainingwhat they might expect the package to do and puttingresponsibility on to the student to find out how to do it.This encourages self-reliance, and a problem-solvingapproach.

As artists and designers we tend to push software andhardware to its limits which means that things can gowrong, so problem-solving is an important talent. Iencourage it.

3.1. TrendsPaul Brown spoke of the future as the invisible placesbeyond an “event horizon” [4]. Extending his analogy Isuggest that we build a platform of current knowledgeand use that vantage point to get a glimpse of whatmight lay ahead.

Programs have become larger as a result of cheapermemory. Developers wish to provide morefunctionality and so there’s an overlap between whatused to be distinct software categories. As a trend thisis good but the consequence is an unweildy andexpensive software package. The concept of ‘plugins’has gained popularity as a way of extending thefunctionality of a software package without making itmonolithic. Many software developers allow for andencourage them. Browsers thrive on third-party pluginsto extend their scope.

Object Oriented Programming is becoming thestandard, and implies a flexibility that will allow forproblems and solutions of which we can’t yet conceiveof. CINEMA 4D is an example of object-orientedanimation software. The manual apologizes at onepoint:

“This is not meant to challenge your imagination butis simply the result of CINEMA 4D being a totallyopen system. Nobody knows the type of animationeffects the future might bring and (what) you maywish to assign them to.”

Web-based Objects are taking off.

4. FUTURE (now to the next Millennium)

4.1. A Disclaimer.Predicting what Art and Design will look like in thefuture is oxymoronic. As soon as you describe it - itexists (in the present.) Witness the failure of Disney’sTomorrowland, and Mr. Spock’s three-dee chessboard. It’s safer to invent a retro-future such asKubrick’s Louis Quinze interior in 2001 A.D.

However we can anticipate a working (and teaching)environment, and possibly even help to make ithappen.

Software packages will cease to exist as we now knowthem. Applications will consist of objects andmessaging. You won’t necessarily know whatcomputer they are on. For any particular task you’ll beassembling (without much effort) the appropriatemodules. You won’t own software any more, you’llrent it, paying a fraction of a penny for each objecteach time you use it. Equally your own objects may besought after items and provide some revenue. Forgetabout software upgrades, or new versions of theoperating system. Any one of the millions of thecomponents of which they consist could be updatedtransparently at any moment in time.

Browsers will probably be the only software on yourcomputer. They will be the portal to your web-basedsoftware modules, and they won’t look like they donow (no clunky windows and that anachronism knownas scrolling).

The result of this will be a trend to tailoredenvironments. As a designer you’ll have your own setof tools (like craftsmen of old) which you will honeand refine. These tools will result from the kind ofwork that you do and will reflect your creativesensibilities. They will be an analog of your distinctstyle.

Think of the potential for teaching. Principles will stillabide, but now you can design a tailored environmentappropriate to who and what you are teaching.

As students mature creatively you’ll teach them how tobuild their own tools (if they haven’t already foundout!) Is this back to programming? Yes, in the sensethat you’ll have control over your own solutions toproblems. But not in the sense of sequentialprogramming with a few simple control structures andvariables. Object-Oriented Programming demands adifferent mental agility and the need to know details ofhuge libraries of existing objects and their syntax.Good user interfaces will facilitate this.

4.2. Output - put out.More of the things designed on the computer will usethe computer as a means of delivery. Physical output,will diminish for several years until the RapidPrototyping Systems become consumer appliances.The Prestigious Prix Ars Electronica Competition hada category called “sculpture” for several years

eventually to replace it with “Web-based media”.Conincidentally this was the year I planned to submit!

4.3. Collaboration - the Good, Bad, and the UglyEarly Siggraph conferences often showcasedcollaborative paint programs. You could add to animage on a monitor at the same time as someone in anadjacent booth or across the world. Somehow theresults were always disappointing (the Bad). Perhaps alack of clear goals lead to what was a competitiverather than collaborative exercise.

Algorithmically generated images have been posted onthe Net which one can rank (the Ugly.) Subsequentgenerations of the art work spawn from images thatwere most popular - Aesthetic Darwinism, perhaps, ormore likely the cyber equivalent of Velvet Elvis.

As part of the Electronic Theater at Siggraph someyears ago, Rachel and Loren Carpenter set up acollaborative visual venture (the Good) whichcontinues to astonish me. Each member of the audiencehad a wand with reflective stickers on each side, greenand red, which they could hold up to signal a binarydecision. The collective audience response wasgrabbed by a camera at the front of the auditoriumprocessed in some way and revealed on the cinemascreen.

In the first exercise a large circle was drawn on thescreen. With the whole screen representing a map ofour seats in the auditorium we were asked to signal onecolor if we thought we were inside the circle, andanother if we were outside. Within a second or so thecircle filled in. In a couple of seconds more the edgesof the circle were well-defined and stable. How didthis happen with such authority? Each person made asingle binary decision. The decision was based onsome knowledge of their location in space. They couldperhaps see what decisions their immediate neighborshad made. The image on the screen provided somegeneral feedback but not specific to an individual’sdecision.

Though in its consequence this example may seemtrivial I think the implications are profound. Thenotion could be extended to creative tasks using theInternet both as the equivalent of the auditorium and ofthe screen. Collaborators would provide far moresophisticated decisions than in our example, and theirrole in the task would be correspondingly significant.

The nature of creative problems that could beaddressed and the design answers we may see areboundless.

References

[1] Paul Goldberger. The Eames Team. In TheNew Yorker, pages 90-99, New York, May1999.

[2] Art Show Catalog, Siggraph ACM, Dallas,1986.

[3] Judy Brown. Teaching Computer Graphics:An Interdisciplinary Approach., pages 189-190 Siggraph ACM, Anaheim, 1987.

[4] Computer Animation Workshop.,CaliforniaState Summer School for the Arts,Humboldt, California, 1990.

Acknowledgements

To Bob Holzman for enabling the Present and Futurein my career and to my Stepfather, Ken Dixon.

Online Education

Anne MumfordAudio Visual Services, Loughborough University

Loughborough, UKa.m.mumford@lboro.ac.uk

Abstract

New technologies for distance learning through the Web, video conferencing,collaborative working tools have the potential to enhance and extend learning. Thisrequires a large leap for many institutions, teachers and students as well as an investmentin technology and training. Many of those involved in graphics and visualization have insome way contributed to the development of the technologies being used and to theirpotential application use. These technologies have potential for widescale use at a timewhen there is pressure, at least in the UK, to widen access to higher education and toconsider learning to be a life-long process.

This paper challenges those involved in Graphics and Visualization Education to look atthe following areas:

Distance LearningFor collaborative ventures to share materials, bring in guest presentations frominternational speakers and to develop and enhance collaborations between those involvedin Graphics and Visualization Education.

Collaborative WorkingResearch has been presented to the Working Group about the possibilities offered bycollaborative working for visualization. The group should promote such work and buildon it widening the application areas and involving other subject areas.

Widening Access – Promoting ParticipationThe technical and pedagogic developments of those involved in Graphics, Visualizationand Virtual Environment education have the potential to open up education to thoseneeding to study remotely from a University. The potential of such techniques should bepromoted within a range of application areas.

Getting Online With More DisciplinesMuch of the development of graphics and visualization tools has occurred withinComputing Science, Engineering and the Physical Sciences. The benefits have greatpotential in other areas, particularly in the social sciences. Data mining and understandinglarge and complex datasets, such as longitudinal datasets, is of considerable importance inthe social sciences. The potential for the use of visualization techniques within researchand teaching seems to be thus far unrealised.

On the Functionality of Distributed Multimedia-System Architecture forEducational Applications

D. Dochev1, R. Pavlov2, R. Yoshinov31Institute of Information Technologies – Bulgarian Academy of Sciences

2Institute of Mathematics and Informatics – Bulgarian Academy of Sciences3Computing Center - Bulgarian Academy of Sciences

dochev@iinf.acad.bg, radoslav@bgcict.acad.bg, yoshinov@bgcict.acad.bg

Abstract

The paper discusses the functionality of distributedmultimedia architectures for educational applications.It reflects some developments made under theinternational project ARCHIMED “AdvancedMultimedia-System Architectures and Applications forEducational Telematics” (part of the EC RTD programINCO-COPERNICUS). The aim of this project is toexplore and implement technological methods andtools for creation and distribution of multimediaeducational materials. The main elements of thearchitecture implementation are the project LocalInternet Support Centres. The functions and structureof the Bulgarian Support Centre are presented with theadopted hardware and software platforms, the initialrepository content and the pilot courses to bedeveloped during the project.Keywords: educational multimedia, distributedarchitectures, virtual campus

1. Introduction

The educational multimedia data management has todeal with several specific features:• The necessity of interactive use of the multimedia

materials, allowing individualised feedback anddiscussion.

• Extensive use of distributed multimedia resourcesavailable on WWW.

• The necessity of effective student navigation toensure a balance between the two extremes of totalconstraint and total absence of constraint,according to the student’s needs and current stateof knowledge.

• The necessity for versatility of composition forfast update and modernisation of educationalcontent.

• The necessity of modular and open-systemorganisation.

• The need of intelligent assistance in informationhandling.

• The availability of a big pool of small educationalcontent providers.

All real implementations of systems, accessingmultimedia documents for a given class of tasks,have to compromise between the requirements forfull functionality and the constraints imposed bythe available resources. Therefore in order to createa viable pragmatically and technologically soundarchitecture the solutions to be applied have toconsider:• the modern tendencies in organisation of

multimedia systems;• the educational needs of the project end-users;• the developments in the modern computer-

supported learning environments;• the current computer and communication

structures of the university partners in theproject and their possible development in thenear future.

2. Conceptual Base of the DistributedMultimedia Educational Architectures

The architectures under discussion are build upon thefollowing types of conceptual models:

1/ Multimedia data models, specifying the structureof multimedia data management systems and themethods applied for storing, composing and retrieval ofmultimedia documents. The multimedia data modelshave to provide useful abstractions and relatedmechanisms to support:

• Generalisation/specialisation hierarchy• Attribute specification for describing theproperties of a document• Specification of operations, performed on amultimedia document• Manipulation of composite objects• Objects sharing• User-oriented presentation.

The multimedia data management requiresmechanisms, common for classical databases to ensureintegration and integrity control, query support,privacy, version control etc. However some featuresare of special interest as due to huge data volumes andvariety of information their achievement is moredifficult - efficient data capture and access; dataavailability; data persistence and recovery; presentationof delay-sensitive objects etc.

2/ Pedagogical model, specifying both the structure ofmultimedia educational documents and various ways oftheir use (for teacher-centered or learner-centeredorganisation of the educational processes). Manycurrent educational projects for distributed computerenvironments apply the principles of constructivistpedagogy and use models, based on the followingleading ideas [1, 2]:• Learning should be context based, i.e. learning

experiences should be contextualised in authenticactivities; learning is acquired through makinglinks with existing knowledge.

• Conceptual learning is through activeinvolvement: a task is understood throughparticipation in it; learning involves creatingpersonal meaning and understanding; experiencebecomes part of the meaning.

• Learning is through collaboration with others:sharing knowledge resolves misunderstandings;interaction leads to new knowledge; understandingevolves from shared knowledge constructing;

• Learner should have personal autonomy andcontrol over learning: modern learning involvespersonal decision making, formulating ownlearning strategies and own goals; teachermediation depends on needs and skills of thelearners.

• Learning is personal growth - argument leading toreflection helps refine concepts; outcomes areunique to the learner;

• Learning outcome is a perspective and anunderstanding: specific content and learningoutcome should not be prescribed; multipleperspectives of the learning task and differentapproaches to understanding are needed.

3/ Model of communications, organizing theinformation exchange of multimedia information indistributed computer environment. The model ofcommunications has to ensure proper multimediaenvironment in networking inside and between thepartners. The partners networks normally includeProxy servers, Cache servers, WWW servers, Nameservers (DNS, WINS and DHCP servers), Mail servers,FTP servers, Terminal servers, Application servers,

Database (SQL) servers. The new quality of thenetwork model is achieved through appropriatemechanisms for cooperation of these servers and theextension of their capabilities using protocols andstandards (e.g. ActiveX, specific plugins etc.)recommended for exchange of multimedia information.

3. End-User Groups and Generic Services

The ARCHIMED architecture is meant to serve twodifferent groups of end-users: A. Learners, accessing and learning withmultimedia educational materials: university students,students in specialised schools, participants invocational training educational forms, non-formal andlife-long learners. The architecture to be developed hasto ensure sufficiently rich environment, relevant to themodern pedagogical approaches and tendencies forteacher-assisted and learner-centered education in localand global computer networks with some collaborativelearning possibilities. B. Authors of multimedia educational materials:university lecturers, high-school teachers, lecturers inspecialised training courses. As a rule the authors arenot IT specialists, so it is not real to expect that theywill be willing to use authoring systems and tools withhigh level of complexity to produce courseware. Themajority of these authors are also meant to use thecreated multimedia educational materials directly intheir teaching practice.

These two groups of end-users require two classes ofgeneric services to be supported by ARCHIMEDarchitecture. These services may be organised using themodern computer-assisted education metaphor of"virtual campus". In order to develop a specificdetailisation of this metaphor, suitable for the needs ofthe project, the participants in the educational processand their main roles has to be revealed. Similarly to [3]we consider the following main actors in theeducational process in the “virtual campus”:

1/ Learner - transfers information intoknowledge. S/he uses the courseware from thedistributed repository (“virtual library”) acting in thefollowing roles: explorer of repository resources;navigator through learning scripts; self-evaluator ofpersonal assignments; participant in collaborativelearning (e.g. peer review, debates in telediscussionsetc.).

2/ Trainer/advisor – assists/advises the learnerin the educational process acting in the following roles:produces diagnosis; advisor; assignment evaluator;coach.

3/ Manager - manages actors and events,acting in the following roles: planner; decision-maker;

supervisor and controller; team or group organiser;director of learning assignment.

4/ Author – supplier of educational content.S/he produces courseware using the MM materialsfrom the repository by means of “author studio” tools,acting in the following roles: document analyst;content expert; teaching materials designer;instructional scriptwriter.

5/ Mediator (“information broker”) –facilitates the navigation of the other participantsthrough the "virtual campus" acting in the followingroles: information communicator; user profilesproduction and maintenance; help in environment use;intelligent help in document retrieval.

Each actor is expected to assume various roles; a givenrole may be assigned to more than one actor. The rolesare related with specific information subprocesses,which have to be supported by the educationalarchitecture under development.

These groups of end-users require two classes ofgeneric services to be supported by ARCHIMEDarchitecture:

A. Services for the learnersIn order to ensure interactive use of multimediacourseware by the learners the ARCHIMEDarchitecture has provide the following components:• Access to a distributed repository (“virtual

library”), containing multimedia documents andinteractive courseware with different levels ofcomplexity;

• Tools, organising the processes of supervisedlearning or self-learning using multimediacourseware in a distributed computer environment;

• Tools, facilitating the navigation through therepository according to the user’s needs andprofile (“information brokers”);

• Tools, organising asynchronous/synchronousgroup educational activities in distributedenvironment (e.g. link to human tutors, bulletinboards, peer reviews, discussion forums,conferencing etc.).

B. Services for the authorsThese services provide means for the authors to store,maintain, archive and finally produce their educationalmultimedia materials. In order to cover this serviceclass the ARCHIMED architecture has to support thefollowing elements:• A document directory system and a database for

storing the digitized pedagogical materials;• A web interface for uploading materials on a web

server;

• A library of documents templates (generic modelsor scenarios) to be used by the author to createpedagogical materials;

• Secure discipline for accessing the web server withthe document templates;

• An instructional methodology to guide the authorin his authoring process.

4. Implementation of ARCHIMEDarchitecture

Though the ARCHIMED architecture doesn’t addressdirectly the whole set of distant learning activities, itcovers an important part of the popular in the last years“virtual campus” metaphor. The virtual campus isstructured in virtual spaces, organising the realisationof the abovementioned information subprocesses. Apossible set of virtual spaces may include [3]:• An information space - repository ("virtual

library") containing various types of documents ordata required by the actors to fulfil their function.It contains multimedia materials (text, graphic,audio and video objects) selected thematically forthe needs of the courses produced/distributed andcourseware on different levels of complexity

• A learning space with the requisite tools tocomplete assignments and to participate tolearning system.

• A communication and collaboration space withtools enabling actors to communicate, carry outgroup activities and participate in remote seminarsand teleconferencing.

• An assistance space where actors may get supportand advice or have their environment customised,by calling upon either online support personnel orcomputer-based help resources.

• An authors' space, containing tools permitting theauthors to store, maintain, archive and finallyproduce their educational multimedia materials.

The models and tools in the architecture underconsideration will be experimented on the followingtarget groups - 1/ graduate students of informationtechnologies studying courses on multimediatechnologies, computer animation, web graphics anddesign; 2/ graduate students of journalism for courseson mass media advertising. The experimentalcourseware is developed considering somerequirements for learner-centered and project-basedlearning.

Considering the current state, tendencies and traditionsfor hardware use in the partner's sites ARCHIMEDarchitecture maintains the existing strong PC

orientation. The supporting operating system andmultimedia database management system have toensure flexibility in application interfaces, systemsintercorporability, connection decisions, networkprotocols, routing, distributed component realization,security on all levels etc. Considering the PCdomination in server and client trends, MicrosoftWindows NT was chosen as an appropriate choice forsystem support. Windows NT server looks properserver operating system for the goals of the project andWindows NT workstation or Windows’95/Windows’98 - a proper client operating system.

For management of multimedia documents in networkenvironment an appropriate MMDBMS seemsMicrosoft SQL Server which has all the capabilities,needed for such types of databases. SQL Server offersdistributed security integrated with Windows NT withcentral control of passwords across a domain and newencryption services for logon and data stream. SQLServer remains an open, interoperable platform thatworks with what customers have and know.

The virtual campus of ARCHIMED relies entirely onthe Internet connectivity for the communications

between the project partners. The respective transferspeeds are function of the communication policies andfinances of the institutions of the project partners.Their networks will be extended by adding Clusterservers for achievement of more power and throughputrunning the applications and Remote Access Serversfor serving remote clients (from universities, schools,SMEs etc.) who will use the developed environment(methodology, tools, templates, courseware etc.) in theframe of ARCHIMED. The connections between theBulgarian partners from Bulgarian Academy ofSciences work at the speed of 100 Mbps, as the threeinstitutions are situated close by. The connections arebased on UTP with Switches (devices like Cisco1900,Cisco 2500, Switch-gateway Cisco3100), using TCP/IPas main routing protocol. The Local Internet SupportCentres use Remote Access Servers, based on PPP andPPTP protocols for support of the remote clients.

More information about the design decisions, functionsand the structure of ARCHIMED architecture may befound in [6, 7]. The overall structure of the BulgarianLocal Internet Support Centre is shown on Figure 1.

Figure 1 Structure of the Bulgarian Local Internet Support Centre

Reference

[1] J. M. Ewing , J. D. Dowling, N. Courtis –Learning Using the World Wide Web: ACollaborative Learning Event. In: Journal ofEducational Multimedia and Hypermedia,Vol. 8, No 1, 1999, pages 3-22.

[2] E. B. Susman – Coperative Learning: AReview of Factors that Increase theEffectiveness of Cooperative Computer-Based Instruction. In:Journal of EducationalComputing Research, Vol. 18, No 4, pages303-322.

[3] F. Henri – Designing a Virtual LearningEnvironment for a Graduate Course onMultimedia. In: Davies G. (ed) Proc. of XVIFIP World Congress, Teleteaching’98 -Distance Learning, Training and Education,Part I, pages 427-444

[4] C. Hmello, M. Guzdal, J. Tums – Computer-Supported Collaborative Learning: Learningto Support Student Engagements. In:Journalof Interactive Learning Research, Vol. 9, No2, 1998, pages 107-129

[5] G. Paquette – Engineering Interactions in aTelelearning System. In: Davies G. (ed)Proc. of XV IFIP World Congress,Teleteaching’98 - Distance Learning, Train-ing and Education, Part I, pages19-41.

[6] D. Dochev, R. Pavlov et al. - Modelling ofData Management. Project ARCHIMEDPL961060, Report D2.1, 1998

[7] R, Yoshinov et al - Modelling of Communi-cations. Project ARCHIMED PL961060,Report D2.2, 1998

Three Collaborative Modelsfor Computer Graphics Education

Colleen CaseSchoolcraft College

Computer Graphics TechnologyLivonia, MI 48152 USA

ccase@schoolcraft.cc.mi.us

Abstract

Visual communication is an area that acts asan integrator to other disciplines, businesswork projects and institutions. Thecollaborative models and innovative deliverysystems described in this document provide“real world” experience, initiate acceleratedlearning and develop flexible structures forcurriculum that meet the needs of business,industry, the learner and the educationalenvironment. These models and samplesaddress many of the current issues facing thecomputer graphics profession.Keywords: collaboration, inter-disciplinary,distance learning, video- conferencing.

1. Introduction

Schoolcraft College is a public, tax-supportedcommunity college serving the people ofnorthwest Wayne County, in Livonia,Michigan USA. The college district iscomposed of five public school districts:Clarenceville, Garden City, Livonia,Northville, Plymouth-Canton and part of theNovi Community Schools. Schoolcraft Collegewas founded October 24, 1961.

The Computer Graphics Technology programat Schoolcraft College prepares students forcareers in computer graphics and graphicdesign through a combination of classroomand real world experiences. With this mix,students learn to apply technical skills anddevelop professional work skills. Thisdocument identifies some issues facing thegraphics profession and describes somestructures and models used to address thoseissues.

2. Computer Graphics Education Issues

• shortage of qualified graphicsprofessionals

• keeping curriculum current with rapidlychanging technologies

• student expectations for high qualityapplicable experience

The shortage of qualified graphics professionalsshould not be over simplified to numbersreflecting supply and demand. The shortage isalso due to a lack of coordinated efforts betweenbusiness/industry and educational systems.Educational institutions struggle to keep thecurriculum current due to both rapidly changingtechnologies and changing industry needs.Students expect and should receive a high qualityeducational experience that is applicable to theneeds of the market. These are not new issues butthe speed of change requires a new way of dealingwith them. The old model of education, workexperience and curriculum development asseparate activities needs to unfold into a real-timesimultaneous activity. Students can learn byexperiencing real projects that simultaneouslyidentify curriculum needs and deliver learning(section 3—Sample).

• complex integrated communications, softwareand hardware technologies require multipleexperts

• faculty development and training are aninstitutional issue

Departmentalized educational institutions do notreflect the integrated multi-disciplined reality ofthe world. Complex integrated communications,software and hardware technologies requiremultiple experts for project/product development.Computer graphics and visual communication is afield that acts as an integrator to other disciplines.Educational systems can provide a value-addedenvironment by allowing interdisciplinaryexperimentation. Coordinated joint instruction isnot just a bonus for the student but has the addedvalue of faculty development. We need to look athow we use our timehow we can allocate

human resources to simultaneously teach andlearn (section 4—Sample).• digital media, and communications

technologies such as Interactive Video-conferencing and the Web aremechanisms for developing newrelationships and experiences incollaborative work

New communications technologies haveopened up opportunities for collaborativerelationships. We have a need to shareinformation in order to remain current in thespecialized technologies we pursue. Largecomplex projects require multiple expertiseand the Internet and Interactive technologiesprovide mechanisms to share expertise andinformation (section 5—Sample).

3. A Flexible Program Structure…enables a “Real World” ProjectPhilosophy

The Computer Graphics Technologydepartment has adopted a “real world” projectphilosophy. To implement this philosophy,Stephen Wroble developed an innovativeflexible curriculum structure that meets thechanging needs of students, faculty, localbusiness and industry, and can adapt to achanging technical environment. By creating atrack system with crossover electives, studentscan create a more specialized sequence ofinstruction. The curriculum structure wasdeveloped with growth and change in mind.

The need for frequent curriculum revisions andfor specialization due to rapid changes ingraphics technology is also built into thecurriculum structure. The design electivestructure provides a mechanism for addingnew courses and a testing ground for newtracks (see Figure 1).

For example, the screen design elective wasimplemented initially for the Multimedia trackand initiated the development of the newestWeb Design track. The storyboarding designelective is useful for many tracks and iscurrently a testing ground for future animationand videography tracks.

There are currently 5 tracks in the curriculum.Publishing, Illustration, Digital Imaging,Multimedia and Web Design. Students choose

2 tracks for an associate degree or can get postassociate certificates (see Figure 2). Our studentpopulation includes both the traditional studentsand graphic professionals returning to update theirskills. This mix gives the classroom an energeticdynamic and a professional maturity.

Figure 1. CGT Tracks

illustration

image processing

publishing

CGT 1102 cr

Freehand

CGT 1132cr

QuarkXpress

CGT 1042cr

Photoshop 1

CGT 1122cr

PageMaker

CGT 1112cr

Illustrator

Design Elective*

2cr

Design Elective*

2cr

Design Elective*

2cr

CGT 2042cr

Photoshop 2

CGT 2204cr

ElectronicImaging

CGT 2304cr

ElectronicPublishing

CGT 2254cr

DigitalImaging

Design Electives Flexible Module ChoicesCGT 150–TypographyCGT 151–Survey of DesignCGT 152–Screen DesignCGT 153–Portfolio DevelopmentCGT 154–Music Sound DesignCGT 155–StoryboardsCGT 156–Photography

Design modules in development–Color Theory–Legal/Business Issues–Interactive Writing–Transformational Geometry–Digital Video

multimedia

CGT 1052cr

Director 1 Design orMusic

Elective*2cr

CGT 2052cr

Director 2CGT 241

4cr

Multimedia

web designCGT 1072cr

Powerpoint

CGT 1082cr

Web Graphics

Design Elective*

2cr

CGT 2354cr

WebDesign

Figure 2. CGT Requirements forAssociate Degree

Each track starts with foundation skills in thebasic drawing and design; application softwareskills follow, along with design electives. Oncethese skills are established, they are applied inspecialized projects for the track (CGT 220,230, 225, 235, 241).

All of the tracks then converge into the CGT250 Practical Applications. Students from alltracks are mixed into this project class. The“real world” project philosophy providesstudents with a career experience in asupportive environment.

In these project classes, students’ partner withbusiness and industry, non-profit organizationsor on-campus marketing design groups toexperience and learn through actual projects.Because all projects do not fit neatly intosemester timeframes, projects can be extendedif necessary into a variety of forms: honorsstudies, internships or additional practicalapplications classes (CGT 251, 270, 298).

As a result of working on a real project,students gain valuable experience in:

• Application of design techniques andprocesses

• Project design and time management• Systems thinking• Budgeting and production constraints and

analysis• Collaboration and teamwork and

relationship building• Decision making based on research and

evaluation• Communications, presentation• Troubleshooting and problem solving• Critical thinking, and data interpretation

Over the past 4 years, the GraphicsDepartment faculty coordinated close to 200large and small projects for non-profit

organizations and the college. These were used inboth the track classes and in the PracticalApplications classes. Examples of the projectsinclude layout design for two literary journals, theMcGuffin and The Michigan Community CollegeJournal. Logo and T-shirt design for the Walk forWorld Hunger and the disabled sports WheelChair Hockey league, building and campus mapsfor the Schoolcraft kiosks and Web site, graphicdesign training materials for Statewide MethodistChurch publishers, displays for the Women’sResource Center, layout and design of theSiggraph Education catalog and CD, and coverdesigns for the Schoolcraft catalog and schedule.

Several projects were also done with six businesspartners. Each business related project wasinitiated through the Workforce DevelopmentOffice. The Workforce Development Office atSchoolcraft College facilitates and implements thedevelopment of partnerships between business,industry, the community and education. Thesepartnerships help contribute to the development ofnew teaching and learning systems for the college.This model allows the business partner to takepart in the learning environment and experiencethe same learning objectives as the student, whileproducing a product needed for the industry. TheProject Driven Learning System and Courses(PDC’s) are an example of a new learning systemfor the college. The first PDC was designed inpartnership with General Motors Delphi.Computer Graphic Technology students produceda series of tools (Job Aids) that were designed tohelp GM employees perform tasks and operations.For the CGT program business partnershipsincluded:

1. GM-DelphiJob Aids and Graphic Design TrainingManuals

2. Ford Motor-EEME Emerging Export MarketEngineersIntranet based time reporting interface design

3. Johnson ControlsCDRom product implementation processtraining

4. CrucamLogo and marketing materials

5. Advanced Communication IncorporatedOperations manual layout

6. Environmental Technology CorporationMarketing Trifold

Imperative in the building of the businesspartnership is the focus on the priority of the

BasicDesign/Draw

3cr

Electives

3cr

CGT 250Practical

Application3cr

Art or CGTElectives

3cr

CGT 270Internship

3cr

CGT 250Practical

Application3cr

CGT 298Honors Studies

3crFirst Track

Second Track

or

or

learning process and not the actual product.Unique instructional tools and methodologieshave also been produced to support thedelivery of PDC’s.

A proposal is written to document thedeliverable product, and also includes theresponsibilities of the learner, facilitator andbusiness partner in the learning process. Theproposal becomes the foundation for thecourse and dictates both the applied technicalobjectives, along with the process. Aframework has also been developed thatidentifies a design process that guides the teamthrough the project. (e.g. Understanding thescope of the project, the business partnersenvironment, needs inventory, team processes,applying skills, devising production plans andimplementing and delivering the product,debrief and learning analysis.)

As the class proceeds the project’s needs arediscovered. This happens through meetings,interviews, surveys, and other needs analysistools. The curriculum needs emerge and aredeveloped around the project needs. In theexample of the Ford EEME interface design,CGI scripting, database and programmingskills were needed along with the graphicdesign and HTML skills that already existed inthe group. A second phase of the project wasestablished as an internship for creating thedatabase connection to the graphical interface.The students also learned about Ford MotorCompany’s culture and global markets.

The students maintain a skill process log wherethey record and reflect on their learning andteam processes. A formal learning analysispresentation is required in addition topresenting the final product.The benefit to the student for all types ofprojects is a real experience in the productionprocess with both instructor facilitators andpartners. The benefit to the facilitator is currentcurriculum and course development that meetsthe needs of actual work. The benefit to thepartner is a learning environment to interactwith potential employees and a deliverableproduct that meets a need. This direct inputinto the educational system benefits both thepartner and the educational institution.Two areas of caution and difficulty inimplementation of the business projects are thepartner’s priority and the project selectionprocess.

Partners who focused on the product and were notconcerned with the learning rushed the processand became a management issue for the facultyfacilitator. Also, projects that were selected moreclosely by the instructors were inherently moresuccessful experiences for the students. Projectsthat were chosen incorrectly became extremelyfrustrating for the faculty. Project selection mustbe driven by the desire to provide the student withappropriate learning opportunities and not bysome other political business relationship need.

By developing flexibility in the curriculumstructure, a program has been developed thatmeets a wider range of individual specializationneeds. By developing a framework forexperimentation with “real world” projects, theprogram can respond to the job market in thecommunity, fulfill non-profit service needs andprovide a resource for the college. Real projectsprovide all participants with a relevant educationalexperience and authentic curriculum development.

4. Interdisciplinary CollaboratingEnvironments …Breaking theBarriers of Departments

Several experiments have been conducted with theComputer Graphics Technology department andother departments at Schoolcraft College. Withthese cross discipline experiments comes aglimpse at the possibility of infinite combinationsof instructional offerings. Most recently in thebusiness arena, three instructors are coordinatingefforts in a course called Marketing and theInternet. This includes a marketing instructor, agraphic design instructor, and a language artsinstructor. The students in this initial offering aremarketing students, but the instructors haveconsidered the integration of students from allthree disciplines. Students are now analyzing Webmarketing from multiple perspectives of languagecontent, aesthetic look, technical and navigationalinformation design, and marketing concepts. Theyhave integrated graphic design and language artsskills into their marketing presentation design,merging different perspectives and blendingconcepts. Faculty are also broadening theirknowledge base and learning from each other.

Other coordinated experiments have included amultimedia class that requires both visual graphicsand the synchronization of sound and a Webdesign project integrating the interface designerand the programmer. An illustration class isproposed to include a graphic design instructor

and a mathematics instructor bringing in thedimension of transformational geometry.

For the student, the benefits of coordinatedteaching are obvious. Coordinated teachingprovides the student with the multipleexpertise of several faculty members orconsultants. The students become aware andunderstand on a multidimensional level. Theyalso witness the collaboration and interactionas the instructors become actively involved inlearning. The students were aware that this wasan experimental environment and providedfeedback and suggestions to all the instructors.This behavior made them more open tocritiques of their own work as they learned toappreciated feedback as an opportunity forimprovement.

A less obvious benefit from joint instruction isthe benefit to the faculty. By workingcollaboratively in teams, around emergingtechnologies, faculty members learn togetherand are able to update their skills; trainingoccurs while simultaneously teaching. Thecourse develops out of the faculty and studentknowledge bases and out of the socialinteraction taking place. This retraining andinvestment in faculty development will pay offin future offerings of the course and in thedevelopment of alternative delivery methodscourseware.

Technology choices were made based on howthe technology enhanced the learning. In theMarketing class, we experimented with email,discussion groups, Web analysis andElectronic Library research along withpresentation technologies. These experimentsquickly brought students together and formed acommunications mechanism to do work.Marketing was the common language of theclass, and the cross-disciplinary approacheswere related to this foundational language. Adecision to use Marketing as the foundationallowed this first offering to be managed andcoordinated around marketing topics, withmarketing students.

Future offerings will allow for betterintegration of language arts and graphic designprinciples into the planning. Future offeringsshould be modified to allow for more facultyintegration and pre-planning time.

Several issues arose from this experimental class.Administrative support is an issue for newexperimental delivery methods. The value of aninnovative faculty member is reflected in thatsupport. This is an issue that we must furtheraddress in the expansion of the course. Investmentin faculty development through joint instructionand support will pay off in the future, allowing forimprovement in the quality of our work. Eachteaming configuration might require a differentamount of overlap. A matrix compensation planwas proposed to assist the administration indealing with the complex issues of compensation.The matrix relates class session percentages byinstructor to a totaled minimum and maximumrange for the course. Factoring in the benefit ofinstructor training can also help justify theexpenses of joint instruction.

5. Interactive Videoconferencing... a Shared Learning Experiencewith Multiple Institutions

The Internet, networks, and interactive video-conferencing systems have expanded theopportunities for collaboration across multipleinstitutions. After attending a NSF workshop onInstructional Computing, staff and instructorsfrom several institutions set out to create acollaborative learning experience for students,instructors and media staff. By combiningcomputer technologies with distance learningdelivery systems, a joint interactive-TV art anddesign project emerged. This collaborativeactivity has been done several times over the past2 years.

Various levels of graphic design students fromthree institutions were given a similar project. Theproject this year was to create a Red Cross BloodDrive poster. Students researched using theInternet and gained insight into blood types, donorpopulations and blood donation needs. SchoolcraftCollege and the University of Wisconsin-Stoutgraphic design instructors used email to preparethe guidelines for the project, while the mediasupport personnel verify the compatibility andconnection of the systems. Schoolcraft Collegeagreed to act as the initiating site.

Second year students from the community collegeparticipated along with fourth year students fromthe University. The projects varied each year buthad some common objectives:

1. Students learned the features and tools ofgraphic packages, while emphasizingdesign principles.

2. Students, instructors and support stafftechnicians were exposed to graphicdesign as applied to video conferencingequipment. This included a look at screendesign and video monitor displayincluding such topics as aspect ratio, safetitle areas, interlacing, NTSC and colorlimitations, along with comparisonsbetween compressed video resolution, TVresolution and computer monitorresolution. This gave the studentsexperience in using features such as colorcorrection techniques, to meet thetechnical limitations of the media.

3. Students had the opportunity to interactwith students from other schools using theInternet and video conferencingequipment. This helped them to developcommunications skills and introducedthem to people who may be futurecollaborators. It also exposed them towork produced by students in otherprograms at other institutions. This isbeneficial to the instructor as both acomparison check and for exposure toother instructors teaching approaches andcritiquing comments.

4. Students were allowed to control theconferencing equipment from the touchcontrol panel, giving them a unique newexperience with a new technology.

Teams of students orally, visually andinteractively presented their work for critiqueusing the interactive classroom. Techniciansassisted in training faculty and students so theycould present live in the Interactive TVclassroom.

The interactive TV room has a British TelecomCodec that does any connection from a singleISDN to a 1/2 T-1. The instructor/student useda control panel that is Windows NT based.Equipment is controlled using a touchscreenicon based panel. This includes audio andvideo switching and peripheral control. Videoand audio routing is through software andinterfaced to a 12x8 video matrix switcher cardand a 4x4 audio matrix switcher. There is anRS232 interface to the professional VCR.

Audio functions from the control panel includemicrophone mute, gain, speaker volume controland mute, VCR volume control and mute, audiocalibration, privacy and refresh. There are twodirectional classroom and instructor microphonesand a speaker systems. Other functions of thesystem include automatic adapting to the acousticenvironment; this eliminates echo, feedback andother noise.

The video functions such as camera selection,camera positioning, AV routing, VCR control andsome testing are also controlled on thetouchscreen panel. There are classroom cameras,an instructor camera, and a document camera;which serves the purpose of an overhead projectoryet allows 3-D objects to be displayed. Twelvepreset camera positions can be stored as a profile,allowing different instructors to preplan theircamera positions. Panning and tilting up down anddiagonally are controlled with variable speed tiedto the firmness of touch. The VCR controls (play,stop, record, pause, fast forward, rewind, counter,reset, eject) are also available from thetouchscreen and video recording can be of thelocal site or the remote classroom.

Other peripherals include; the Computer withInternet access, a video disk player, a facsimilemachine, telephones for fax, voice-speaker, andremote diagnostics. There also is the capability totake control of the remotes site’s cameras. Oneremote site demonstrated their wireless trackingsystem feature. The tracking system allowed theinstructor to move around the room and remain oncamera.

Schoolcraft College belongs to MiCTA. This is acommunications organization that follows therecommendations of the (ITU) InternationalTelecommunications Union. The ITU coordinatesglobal telecommunications networks.(www.itu.int) MiCTA is a support mechanism forthe college technicians and works to maintaininteroperability specification for equipment andconnections with other sites. Information aboutsimilar video conferencing systems can be foundat (www.micta.org/mictamap/).

Students from Schoolcraft College andWashtenaw Community College and theUniversity of Wisconsin-Stout have successfullyparticipated in joint project critique sessions usingthe two way audio and video - computerinteractive television classroom.

This project could be expanded. SchoolcraftCollege has agreed to act as the initiating site.The goals of this expanded project couldinclude the following:

1. Create a 4-way link using a bridge toconnect multiple schools.

2. Experiment with partners from other areasof the world. Initially in point to pointconnections and then in bridgeconnections.

3. Initiate student project teams frommultiple institutions to collaborate anddevelop relationships across geographicboundaries, using Internet emailcommunications with a few sessions forcritique and sharing in the Interactiveclassroom environment.

An Internet variation of this concept could bedeveloped for those who do not haveinteractive video capabilities.

6. Summary

The benefits of the collaborative modelsarticulated in this paper deal with some of theissues and challenges of educating students ina rapid changing environment. They do nothowever come without some problems inimplementation. As technologies continue toevolve and change at incredible rates,education must look at ways to integrate thetasks of education, work experience andcurriculum development. The CGT programstructure is very effective in providing adynamic curriculum model. The program hasgrown from 2 to 5 tracks in less than 4 yearsand has doubled in enrollment. Employer’shiring CGT graduates have sent positivefeedback to the program. Students expect ahigh quality applicable individualizedexperience. The flexible track and modulardesign elective structure allows students totailor their curriculum to meet their uniqueinterests or needs. Some students although,require extra guidance in this task, havingexperienced more rigid, less fluid structures intheir past. Counselors also have difficulty withthe choices and rapid changes in thecurriculum. Faculty find themselves taking onan advisory role. Curriculum developed aroundactual real project requirements and

employment needs has helped identify weaknessesand needs in the college’s curriculum and hasaffirmed some of its strengths.

Faculty development must also occur quickly in arapidly changing world. Joint coordinatedinstruction can promote peer learning amongfaculty. The course exceeded its goals for theinitial offering, and identified some issues forimprovement. A more integrated articulation oflearning objectives will help coordinate effortsand assessment. The benefits to the future of thecollege along with the cost of faculty retrainingmust be part of the equation, when funding is re-evaluated.

We now live in a networked environment thatprovides us with wonderful opportunities forconnections that we did not easily have in thepast. Faculty from multiple institutions haveopportunities to compare outcomes andcollaborate in learning projects usingcommunications technologies, such as InteractiveTV and the Web. Opportunities exist to build andmaintain global relationships through experiencesenhanced through technologies.

Computer graphics as visual communication canact as an integrator for interdisciplinary courses,cross-institutional experiences, and “real world”projects. Improvement of learning happens withfrequent interactions and active involvement, thisis required in collaborative learning environments.Collaborations are waiting to be imagined.

Tell me and I’ll forget.Show me and I may not remember.Involve me, and I’ll understand.Native American saying

Acknowledgements

My thanks to the GVE’99 workshop for theopportunity to initiate dialog on the modelspresented in this document.

My appreciation to Jeanne Bonner, FernonFeenstra, Sam Gooden, Mark Harris, Susan Hunt,Lisa Jacobs, Ray Krawczyk, Susan Lupo, MichaelMehall, John Miativich, Lori Navalta, ChristineRejniak, Don Ryktarsyk, Faye Schuett, RobertSmiley, Gayle Stanek, Bonnie Taylor, BrettWilson, Gordon Wilson, Frank Wiltrakis, and allthe instructors and students who shared in thecollaborative efforts described in this document.

My thanks to Schoolcraft College, UW-Stout,Washtenaw Community College, GM-Delphi,Ford Motor-EEME, Crucam, ETC., JohnsonControls, Advanced CommunicationsCorporation, and all the non-profitorganizations for their willingness to

experiment with new teaching and learningmodels.

A special thanks in particular to Stephen Wroblefor his patience, insight and vision in the creationof the flexible structure used in our program.

Enabling Educational Collaboration - a New Shared Reality

Judith R. BrownManager, Advanced Research Computing and Visualization

The University of Iowajudy-brown@uiowa.edu

1. Educational collaborations, using enhancedcomputer graphics, high performance networks,and virtual reality technologies, will enable newand exciting educational opportunities. Tele-immersive collaborations can provide new learningexperiences and insights by joining educators andstudents from remote locations in a shared virtualspace. Bringing together students with culturaldifferences and diverse perceptions of reality intothis shared immersive space allows them to see andhear the same objects and sounds and to share theirown unique views and perceptions, creating a new,shared reality. My vision of this shared reality forlearning is based on three premises regardingresearch and learning.

2. Premise 1: Methods developed forresearch naturally find their way into theclassroom.

The visualization lab has long been the crossroadsbetween research and education. The researchcomponent in a university is an important facet ofthe educational environment and cannot be viewedas separate from "education." Research is dependentupon student assistants, both undergraduate andgraduate, many of whose theses will grow out ofthe research. A professor's research pushes studentassistants' learning experiences, and the resultsfrom the research by both faculty and students findtheir way into national presentations and into theclassroom.

Besides the fact that research labs provide theeducational environment for student assistants, theclassroom needs for computer graphics will alsopush technological limits since student labs andclassrooms need the same capabilities as researchlabs. Interaction is essential for collaboration, andthe appropriate user interface is important, lest thecomputer get in the way of the students'exploration and interfere with students' naturalcuriosity.

3. Premise 2: Virtual reality, as aninteractive technology, is a valuable toolfor instruction.

As virtual reality technologies have recentlyenabled scientists to better understand their data byexploring the images of the data, so are thesetechnologies enabling students to better understandthe concepts they are learning. Although theseconcepts may not be new to the world, they arenew to the students, and the same joy of discoverytakes place.

The University of Iowa is making progress withapplications on three ImmersaDesks acquiredthrough National Science Foundation funding foreducational purposes and environmental research. Ayear ago, there was very little software forImmersaDesks, and our programmers had a longlearning curve to get started. Now some good toolshave been developed and are offered free to the usercommunity. These tools include CAVE5D, aversion of University of Wisconsin's VIS5Dsoftware for environmental visualization, andLimbo, from Unversity of Illinois at Chicago thatserves as a template for developing tele-immersiveapplications, providing avatars, pointers, sound,and other tools for working together in a sharedimmersive environment.

We have ImmersaDesk applications in higher-dimensional mathematics, geography, chemistry,biochemistry, and computational fluid dynamics.Some applications have begun as research andmoved to the classroom. Some entire classes comeover to the ImmersaDesk, so the students canexperience and share the instructor's discoveries. Inother cases, graduate students are sent over to workwith us.

One of the key projects being developed with theNSF funds is the ENVISAGE Project( En vironmental Vis ualization a nd G eographicE xploration.) The purpose of ENVISAGE is toallow students to experience a more completeunderstanding of geographical processes by having

them operate in the semi-immersive 3Dvisualization environment provided by theImmersaDesk and high performance networks.Immersion, for analysis of mapped data and modeloutput, enables students to view relationships thatwould otherwise remain shrouded by thelimitations of widely used graphics technologies.Openshaw and Fisher, for example, assert thefollowing:

“It is important, therefore, to developtechnologies that attempt to understand thedata and develop views of the information ordatabase world in which attempts are madeby our clumsy spatial analysis tools to findpatterns and relationships. Can we lookinside spatial databases and walk around inthem via virtual reality concepts? Currentlywe are so blind to many aspects of the dataand the data flows being analyzed. We copeby being highly selective and subjective andin the process probably fail to see and findmany of the patterns and relationships ofpotential interest other than those weblindly stumble over by chance!”

It is precisely this type of immersive exploratorylearning that we are bringing to undergraduateeducation so that students will be able to morereadily grasp basic concepts that have a pronouncedvisual component

4. Premise 3: Enhanced computergraphics and tools for interaction, alongwith high performance networks, enable ashared learning environment.

We have learned a great deal in recent months byparticipating in tele-immersive applications ofothers and are now developing our applications.Just as early research applications made their wayinto industry, so have they also emerged as toolsfor instruction.

Networked, collaborative learning can begin at avery young age, as shown by the NarrativeImmersive Constructionist/CollaborativeEnvironments (NICE) project between theElectronic Visualization Laboratory and theInteractive Computing Environments Laboratory atthe University of Illinois at Chicago. In NICE,children can collaborate to plant and cultivate avirtual garden and to create stories from theirinteractions. See Figure 1.

Figure 1. Boy in the NICE garden. Imagecourtesy of the Electronic VisualizationLaboratory, University of Illinois at Chicago,1997. The NICE Project: Narrative,Immersive, Constructionist/CollaborativeEnvironments for Learning in Virtual Reality,developed by Maria Roussos, Andrew Johnson,Jason Leigh, Craig Barnes, Chris Vasilakis,and Tom Moher.

As universities have connected to the very highperformance backbone network service (vBNS),they have looked for high-profile applications toshow faculty and administrators the value of thesenew educational and research networks.Applications that show an ability to link withother universities and explore together in sharedvirtual space have been popular. As part of thePennsylvania State University Internet2 Days, weexplored a model of the city of Berlin with anadditional proposed building designed by a PennState professor of architecture, as shown in Figure2. Boston University produced a wonderful,whimsical Art World, and we appeared as avatars toplay in that world during the Supercomputing 98conference. One of the 3D models from Art Worldis shown in Figure 3.

What we learned from these experiences, we cannow apply to our applications and build newcollaborations worldwide. We recently worked withthe National High Performance Computing Centerof Taiwan. They were celebrating the connection oftheir national high performance network, TAnet, tothe vBNS and other national and international highperformance networks. Because the time for theirplanned celebration was 3 a.m. our time, theyrecorded our tele-immersive session earlier. Forthis presentation, we investigated together thepressure fields on a high speed train going througha tunnel, shown in Figure 4. We are currently

working on geographic and computational fluiddynamics applications, and we hope to developinteresting collaborative opportunities with otherparticipants at GVE99.

Figure 2. CAD model of Berlin with sculpturebuilding as proposed by Lebbeus Woods.Modeled by Reggie Aviles, Pennsylvania StateUniversity (Penn State) Architecture Student..Translated for use on the ImmersaDesk by RayMasters, Center for Academic Computing andAffiliate Associate Professor of Architecture,and George Otto, Manager of the Center forAcademic Computing Visualization Group atPenn State.

Figure 3. Image of the Martian Landscapefrom Art World, courtesy of the ScientificComputing and Visualization Group at BostonUniversity. The following artists contributed tothe scene: Matt Harter, Paul Haman, TomCoffin, Jeong-Hoon Lee, and Karlo Takki.

Figure 4. Image of the ImmersaDesk showingthe pressure fields on a high speed train as itgoes through a tunnel, the avatar of the remotecollaborator, and the pointer of the Universityof Iowa participant. This image illustrates atele-immersive application between theNational Center for High-PerformanceComputing of Taiwan (NCHC) and AdvancedResearch Computing Services andVisualization, Information TechnologyServices, The University of Iowa.

There are still some issues with wide spread use ofthese collaborative technologies:1. Cost - Shared immersive technology is

expensive2. Accessability - The size of the equipment

requires large spaces. The interfaces are notreadily accessible by persons with disabilities.

3. Distance - Time zone differences make same-time collaborations difficult. in the Taiwanexample, we settled on 6 p.m. our time on aTuesday, 8 a.m. Wednesday in Taiwan.

4. Network performance - Complicated graphicsand sound require fast networks and lowlatency. In the Taiwan experiment, thegraphics were good, but the six second timelag for the sound was disconcerting.

I look forward to meeting with other educators atGVE 99 to discuss these issues and culturalroadblocks to tele-immersive educationalapplications. I expect to be able to form someinteresting collaborations through which we canshare our tools and knowledge and make good useof the national and international high performancenetworks for educational advancement.

References:

[1] Brown, Judith R. and SteveCunningham"Visualization in HigherEducation," Academic Computing, March,1990.

[2] Cunningham, Steve, Judith R. Brown, andMike McGrath, "Visualization in Scienceand Engineering Education," IEEE Tutorial:Visualization in Scientific Computing,Gregory M. Nielson and Bruce Shriver Eds.IEEE Press, 1990.

[3] Brown, Judith R., "The multi-facetedblackboard: the future of visualization foreducation, The Impact of ComputerGraphics on Education: the Challenge ofInteractive Learning, S. Cunningham and R.Hubbold (Eds.), Springer-Verlag,Heidelberg, 1992.

[4] Openshaw, S. and Fisher, M.M. 1995. Aframework for research on spatial analysisrelevant to geo-statistical informationsystems in Europe. Geographical Systems,2 (4):325-338.

[5] Brown, Judith R., "Computer GraphicsEducation and Presentation" Presentation atIGD opening, October, 1998.

This work is supported by the National ScienceFoundation, Division of Undergraduate Education,under the grant DUE-9750537 and by the NationalScience Foundation, Advanced NetworkInfrastructure, under the grant NR-9613871. Allopinions, findings, conclusions andrecommendations in any material resulting fromthis work are those of the principal investigators,and do not necessarily reflect the views of theNational Science Foundation.

Supporting Face to F ace Teaching

Anne MumfordAudio Visual Services, Loughborough University

Loughborough, UKa.m.mumford@lboro.ac.uk

Abstract

The needs of Graphics and Visualization Education(GVE) are often more demanding of teaching facilitiesthan are other subject areas. In the UK there ispolitical pressure to improve the student experiencethrough assessment of subject areas and as a result ofthe introduction of student fees. One area is the need toimprove the teaching facilities and methods of delivery.GVE may be considered to be very specialist and not akey driver for facility improvement. The requirements,when met, can however provide improvement for all.There is also a need for widening the use of thetechniques. Despite many awareness raising exercises,the techniques are still seen as being specialists andthere is little penetration in some subject areas.Keywords: teaching rooms, teaching methods, onlinelearning, student experience.

1. Introduction

When delivering Graphics and Visualization Educationthere is a need for the provision of appropriate teachingspace and/or technology for flexible learning bystudents at a time and place of their choosing.

This paper focuses on the needs of the teacher and thestudent for face-to-face teaching and the pressures onHigher Education Institutions for the provision ofsuitable teaching space. The perspective given in thepaper is a UK one and is based on work undertaken bythe Advisory Group On Computer Graphics (AGOCG[1]) and the funders of that initiative, the JointInformation Systems Committee of the HigherEducation Funding Councils (JISC [2]). Some of thiswork is now being taken forward by the UK’s StandingConference for Heads of Media Services [3].

The paper concludes by recognising the benefits whichmight accrue to many teachers and students if therequirements of GVE are met within teaching spaceprovision,

2. Teaching Space Provision

Provision of teaching space in UK Higher EducationInstitutions tends to be a mix of centrally owned rooms

and departmentally owned rooms with supportprovided by the centre or by the department.Equipment in rooms may be installed or may besupplied as required for particular lectures, normallythrough a booking system. The result of centrallyowned space and insufficient funding is that use of thespace is efficient but that the facilities available and thesupport provided tends to be at a relatively low levelacross all rooms. The facilities are often inconsistent.Booking of rooms is rarely based on facilities requiredby a teacher but simply on student numbers and,perhaps, special access requirements, e.g. wheelchairaccess. The AGOCG case studies showed virtually nolinks between room bookings and the requirements ofthe teacher for facilities to support their preferredteaching method. The result is a very slow movetowards the use computers for face to face teaching.Lecturers tend to prefer to use low levels of technologye.g. OHPs, slides and writing boards as these can beguaranteed.

3. Political Pressures

There are a number of political pressures which makethis an interesting time in UK Higher Education.

3.1. Teaching vs ResearchThe UK Higher Education Institutions are all subject toexternal reviews by Government. These cover research,through the Research Assessment Exercise andteaching and learning through the External SubjectReviews (formerly the Teaching Quality Assessment).

The pressure to deliver results in both areas is commonto all institutions (about 185 in the UK), though aimsand aspirations vary. The pull on resources asinstitutions strive to improve on research results canhave an impact on the funding for teaching resourcesand support.

3.2. External Subject ReviewsThe External Subject Reviews focus on the studentexperience and are an intensive assessment of thedelivery of teaching and the learning within a subjectin a University. The provision of appropriate teachingspace and facilities is part of this review.

3.3. Flexible LearningThe current government strategy to develop life longlearning and to expand the Higher Education sectormeans that a greater range of teaching and learningmethods need to be put in place if this is to beachieved. The need to provide more flexible learningmethods allowing students to learn at a time and placeof their own choosing for some or perhaps most oftheir studies will be embraced as a concept in differentways by different institutions. Many will adopt newmethods to some degree.

3.4. Student FeesA further pressure to focus on the student experiencecomes from the very recent introduction of student feesto be paid by all in UK Higher Education.

4. The Perspective of the Teacher

The teacher needs to feel that they have a guaranteedavailability of facilities before using differenttechnology for their teaching. The lack of online accessand data projection in teaching rooms results in cautionand a lack of willingness to use new technology inteaching face to face situations. There is also a lack ofconfidence in many teachers which needs to beovercome with training and support. Teachers need tothink about fall-back procedures if things go wrong andto be supported to develop these.

5. The Perspective of the Student

Students are expecting to be able to access resourcesonline. Lecture notes and support materials should bemade available online to support the wide range ofstudents enjoying higher education today. Suchstudents may be working their way through University(common in the US but a new feature of UK HigherEducation) who may not be able to attend all lectures.The availability of materials online is also critical insupporting the education of some students withdisabilities who may only be able to access materials ina way modified to support their particular disability.

Students also expect to be able to see the materials andmany rooms are unfortunately not suitable for dataprojection. Some examples and good practiceguidelines are given in the AGOCG case studies.

6. Teaching Space fit for GVE

Teaching space that is “fit for purpose” is the key to asuccessful experience for both the student and theteacher. The requirements for GVE are towards the topof the range of requirements. Online computer access,data projection, good resolution display, suitableblackout, dimming lights are all needed. Suchrequirements would benefit all. Yet, these facilities arefar from common in UK Higher Education across thelarge numbers of teaching rooms.

7. Widening the Use of GVE and Promotingthe Benefits

At the same time, there is a lack of both awareness andtakeup of graphics and visualization tools in manydisciplines. Many disciplines could benefit from theuse of these tools in both research and in teaching. Inmany disciplines the lack of takeup has been a result ofa number of factors including lack of inter disciplinaryworking and a lack of equipment funding.

The social sciences have been identified by AGOCG asdisciplines which could benefit from the takeup ofvisualization techniques, particularly in regard to theanalysis of large and complex datasets and other datamining activities. Workshop output and case studiescan be seen on the AGOCG Web site.

There is a major contribution still to be made by thosewho have been involved in Graphics and Visualizationtechniques in both research and teaching for someyears. The techniques are applicable to many subjectareas which have thus far still been little touched byawareness of the potential of these techniques. The useof such techniques in a wider range of subjects and thedemand that would inevitably cause for improvedteaching facilities and the development of newteaching methods in areas beyond Computer Sciencewould benefit many.

References

[1] Advisory Group On Computer Graphics,http://www.agocg.ac.uk (including variousreports relating to Lecture Room Services at:http://www.agocg.ac.uk/reports/mmedia/casestdy/casestud.htm

[2] Lecture Room Services, Workshop Reporthttp://www.jisc.ac.uk/pub98/assist3.html

[3] Standing Conference For Heads of MediaServiceshttp://pari.mmu.ac.uk/msh/welcome.htm

Web Based Multi-Us er Interactive Graphics Worlds for Education

G. Scott OwenDepartment of Computer Science

Georgia State UniversityAtlanta, Georgia, USA

owen@siggraph.org

Abstract

Several technologies are currently beingdeveloped that will have a huge impact on thefuture of education. Advances in computationalpower and graphics capability will give the futurehome PC the capabilities of very high-endworkstations of the present. Advances inconnectivity will allow everybody to be connectedat all times with a very high bandwidth. Advancesin computer graphics software technology,combined with advances in other fields, such assimulation and artificial life, will allow us tocreate realistic worlds for students to investigate.These worlds will be Multi-User SharedEnvironments (MUSE), i.e., students will be ableto interact with instructors, other students, andintelligent agents in these worlds.

In this paper, I will briefly discuss the backgroundissues, vision for the future with some specificexamples, and our current efforts in this area.

1. Background Issues

1.1. TechnologyCurrent home computers in the $2,000 price rangehave a substantial amount of graphics andprocessing power, e.g., a 400 Mhz P II, 128Mbytes Ram, Open GL accelerated graphics cardwith 8Mbytes of RAM, a 12 Gbyte disk, and a 17-inch monitor. This is rapidly increasing and withina year this class of machine will have a 700Mhzprocessor with enhanced instructions for 3DGraphics and MPEG-2 decoding, e.g., the IntelKatmai cpu, 256 Mbytes RAM, a 20 Gbyte harddisk drive, and a faster graphics card.

If we assume that Moore’s law (computers doublein power approximately every eighteen months)will continue to hold, then a conservativeprediction for the home PC (with a cost of under$2,000 U.S.) of the year 2009 will be as follows:

• Processing power about one hundredtimes that of a Pentium 400 mhz cpu

• Main memory of about one to twogigabytes

• Disk space of about one terabyte

People will be able to carry these computersaround and use portable displays such aseyeglasses, i.e., the computing will be ubiquitous.Another technology advance is in improvedInternet bandwidth into the home. Currently mostpeople connect via 56 K modems, but more othertechnologies are rapidly becoming available. Forexample, in many cities either Cable Modems ortelephone based DSL lines are becoming availablefor about $50/month. These allow for speeds up to1.5 mbits/sec or about as fast as a T-1 line.

By 2007 this will be of the order of fast ethernet(100 mbytes/sec) and will be wireless. Theintroduction of VRML 97 and Java 3D, combinedwith the rapid advances being made in thesimulation of animals (artificial life) and ofhumans are allowing us to approach the ideal ofthe Star Trek Holadeck. We will be able to createworlds for students to inhabit such that they canchange the parameters or else just live with othercreatures.

1.2. Trends in EducationThree major trends in education are:1. The increasing emphasis on Distance

Learning, especially Web based learning2. Life-long education3. The increased use of technology in education

This means that web based learning technologiesand interactions are becoming increasinglyimportant. Students, especially older students, areless and less interested in coming to a centralcampus to listen to traditional lectures. They wantto be able to learn on their own time and at home.However, there is still a role for the professor-student interaction, either one on one or with oneprofessor and a few students, although this maybecome a virtual interaction.

2. Vision for the Future

Imagine an Art History class where the instructoris holding office hours in the class world, which isstructured as a virtual Art museum. A studentapproaches the instructor avatar and states that shehas a question on a particular painting. Theinstructor and student walk to the painting and theinstructor answers the student’s questions. Theinstructor goes into another room and finds threestudents discussing a particular sculpture. Theinstructor joins in the discussion and walks thestudents around the sculpture, pointing outdifferent aspects of the work. Note that theparticipants are at their homes and are interactingvia computer and conversing using microphones.

If a student logs into the art world and theinstructor is not there, then the student can askquestions of her virtual companion. Thecompanion, one for each student, is an intelligentagent that knows some information about the artworks in the museum. But, perhaps moreimportantly, each companion is privy to thesessions between the instructor and the otherstudents, and so can report about new interestinginsights that the instructor has given to otherstudents.

Imagine a biology class where the instructor leadsa group of students on a tour of a cell. One of thestudents asks about the DNA replication processand the instructor leads the group into anotherworld where this is occurring. The instructordiscusses what is happening with the studentswhile they observe the process. The instructorand/or students are able to modify the process bychanging certain kinetic parameters and observethe changes.

Imagine a zoology class where the students canenter a virtual ocean and observe the interaction ofdifferent species. The students might themselvesbe different types of aquatic creatures, one mightbe a shark and others might be a dolphin,swordfish, whale, etc. They can design a newspecies of predator, set up its parameters andobserve how it changes the inherent ecology.

All of these scenarios involve the use of a 3DGraphical MUSE. In some of the environments,e.g., the Art museum, the primary interactionwould be between students and their virtualcompanion and the professor or other students. Inother environments, e.g., the virtual ocean, thereare other entities that interact with the students,

i.e., the students might not know which of theother creatures were other students and whichwere virtual creatures that were programmed tobehave in a realistic fashion.

3. Current Work

While the above scenarios are futuristic, we havebegun to work to develop aspects of them. We arecurrently working with Art History faculty fromthe School of Art and Design at Georgia StateUniversity to design and implement a prototype ofa MUSE Virtual Art Museum. Our goal is to createa generic museum that can be populated withdifferent sets of art works depending upon theparticular course. For example, a course coveringRenaissance Art would require one set of workswhile a course focusing on the Twentieth Centurywould require a different set.

We are using the Sony Community Place serverfor this project and currently have a demo versionthat allows three simultaneous users (we arenegotiating for a system that will allow up to fiftysimultaneous users). With this system, eachstudent can be given a CD-ROM with that containsthe Virtual Art Museum and a set of avatars. Eachstudent will be assigned to be a particular avatar.The student can log into the server, and interactwith the other students who are logged in.Currently this interaction can either be via typedtext or microphone. An advantage to this system isthat the bandwidth requirements are kept low.Each user has the basic world plus avatars on theirown machine so all the system needs to transmit isthe updated viewpoint of their own avatar, theupdated positions of the other avatars, and anycommunication (via text or audio). Therefore,students at home with only modem connections tothe Internet can use it.

This project has just begun and we will report onour progress at the GVE 99 workshop. Our initialplans are to implement the Virtual Art Museum,test it out with GSU students, and then to beginimplementing some aspects of the intelligentagents.

Conclusion

The convergence of rapidly improving technology,especially the improving graphics capabilities ofinexpensive computers, plus the increased need fornon-traditional distance learning has given rise tothe possibility of using web based interactive 3Dgraphics worlds to radically improve education.

These worlds will be multi-user so they will beable to facilitate student-student and student-instructor interaction at a distance. We have begunwork on the design and implementation of onesuch world, to teach Art History courses.

Using Web-Based Computer Graphics to Teach Surgery

Ken Brodlie Nuha El-Khalili

Ying Li School of Computer Studies

University of Leeds

Position Paper for GVE99, Coimbra, Portugal

Surgical Training

Surgical training is a long and expensive process. Much of it is done by mentoring, where anexperienced surgeon trains a junior colleague - either with the help of models, or on real patients. Thisis demanding on the time of experienced surgeons, and potentially hazardous for patients. Recentlythere has been considerable interest in the use of virtual reality approaches to surgical training: thesehave the potential to allow trainees to work at their own speed, in a fairly realistic way and in a safeenvironment.

Virtual Environments for Surgical Training

However typical virtual environments for surgical training are:

dedicated installations using custom-built apparatus, which can only be used in one location expensive - often based on Silicon Graphics high-end products designed for individual rather than large class teaching specific to one type of surgery in prototype form as a research exercise rather than in everyday use for training lacking in serious evaluation as to the positive transfer of training skills

We have been investigating a different approach. Can we create simple but effective surgical trainingapplications that can be delivered over the Web? These will be VRML worlds which will executewithin a Web browser. If we can do this, then the advantages are considerable:

accessibility: the application can be run from anywhere in the world low-cost: the application can be run from a PC without specialist software (simply a VRMLplug-in to a Web browser) distributed : if powerful computation is required, this could be provided on a remote server, sharedbetween all users class size: large numbers of trainees can be handled generality: one could imagine a family of applications, with a consistent methodology, beingmade available for a range of surgical procedures

Of course - there is a price to pay, namely realism. Can we provide sufficient realism in a Web-basedsimulation in order to provide useful training?

Applications

We have been collaborating with three groups of surgeons and radiologists at Leeds, in order toinvestigate the potential of Web-based training. All three applications - while coming from quitedifferent branches of surgery (vascular, neurosurgery, orthopaedic) - require the surgeon to manipulate atool through the human body. Our approach likewise is common: we convert data from a body scannerinto a geometric representation of the part of the body involved, storing this geometry as VRMLprimitives; we create a geometric model of the surgical tool, again in VRML; and finally we providesome virtual interaction where the user can manipulate the tool through the body. Two of theseapplications are described below.

Neurosurgery

Neurosurgery is a very demanding area of surgery where complex operations are required. These areoften composed of a sequence of simpler procedures, with critical decision-taking required in putting thesequence together. If things go wrong, they go wrong quickly and disastrously. This makes it anexcellent application area for web-based virtual environment training: the simple procedures can berehearsed safely, efficiently and inexpensively.

We have developed one simulator for the treatment of intractable facial pain (which if untreated can beso unbearable as to result in suicide). An accepted treatment is Percutaneous Rhizotomy: it is a safe

procedure when performed by experienced hands. Surgeons direct a needle into the medial portion ofthe foramen ovale, to partially destroy the sensory root for relieving the facial pain. In practice,surgeons mark three points on the patient’s face: from these points, a surgeon envisages the coronal andsagittal planes through two of the points, and directs the needle from the third point to the intersection ofthese two planes.

The simulator is shown above. The application is a linked combination of VRML worlds. The skull ismodelled in terms of VRML (a large IndexedFaceSet) and the needle too is a VRML object. The application allows the trainee to mark points on the skull, andinsert the needle through the skin, manoeuvring it through a hole in the skull. The coronal and sagittalplanes can be highlighted if the trainee needs some guidance (see the picture above). We have usedthree VRML worlds. One world acts as the control panel, with manipulators that simulate handoperations of the surgeon. The other two give views of the patient: the large view in the lower part ofthe application is an ’exocentric’ view from outside; the top-left gives a more ’egocentric’ view, as seenfrom the tip of the needle. Collision detection is important - but a problem in VRML which supportsonly viewer-object collision, not inter-object collision which is what is required here. However, we canuse the egocentric view to capture collisions, and transmit collision information to all the linked worlds. The linking of the VRML worlds is carried out using the External Authoring Interface (EAI).

We have developed another simulator for craniotomy where part of the skull is removed so that theneurosurgeon can gain access to a brain tumour. In this procedure, the neurosurgeon drills a number ofholes in the skull around the area to be removed, and then using a small saw, makes cuts between theholes.

This simulator is shown below. The application is a VRML world with a model of a human skull as alarge IndexedFaceSet, with a brain model inside. The trainee can make the skull transparent in order tosee the tumour. In the training exercise, the trainee marks points corresponding to the holes to beremoved. The simulation then removes the triangles within the area to be cut out, so that the trainee cancheck the success of the virtual operation.

Vascular Surgery

Interventional radiology is another complex area where web-based virtual reality has the potential tobecome an important training aid. In our work we have been looking at a simulation of treatment ofabdominal aortic aneurysms. This is a swelling of the aorta; if untreated, the aneurysm may burstcausing a critical condition. The treatment involves placing a stent inside the artery to shield it from theanuerysm. This is achieved by inserting a series of guidewires and catheters from an incision in thegroin. These instruments are manipulated through the femoral artery, and into the aorta. It is a highlyskilled operation, requiring some ten years’ training.

The simulation is shown above. It is altogether more complex than the neurosurgical applicationsbecause we need to model the deformation of the catheter or guidewire as it collides with the walls ofthe artery. Again we have a VRML world which contains a model of the aorta and the surgicalinstruments. Top left shows the exocentric, overall view, top right the egocentric view from thecatheter. The lower left area shows an interactor that will push/pull, or twist, the catheter. The lowerright shows a CT scan corresponding to a 2D slice through the aorta which can be selected by thetrainee.

The system architecture for this application is quite complex. Deformable modelling requirescompute-intensive calculation - which may be beyond the capability of the client machine. Thus thisapplication uses a server process to execute the physically-based modelling code which computes thedeformation of catheter and guidewire. It also tests for collision between the catheter and artery wall. Communication between client and server needs to be as efficient as possible, since the combinedapplication needs to operate in real-time.

Critical Issues - Interaction, Visual Realism and Haptic Feedback

This is very much work in progress but we now have three years experience and are able to identify thecritical issues.

In our approach we rely only on a mouse as physical input device. Mouse movements are mapped tomanipulation of virtual input devices in the VRML world - in the neurosurgery application, to simple,generic devices; in the vascular application to a virtual representation of the real catheter. The issue fortraining is to what extent this can replace the real input device.

We also need to understand the extent to which visual realism is important. In the vascular case, theradiologist has only a mental 3D model of the vascular structure - there is no camera on the end of thecatheter. This mental model is constructed from intermittent X-ray images, and from feel - plus years ofexperience.

Finally, in both applications, haptic feedback is important in the real-life simulation. An expertradiologist can operate almost by touch alone. With simple Web-based tools we cannot provide forcefeedback (although force feedback mice are soon to be available). However in the vascular case, wehave been able to simulate force feedback by colouring the virtual catheter.

Conclusions

We are studying whether simple Web-based computer graphics can help in the training of surgeons andradiologists. We can never hope to achieve the realism of expensive, dedicated VR simulators, but webelieve we can create useful tools nonetheless. Through VRML we gain a solution that is both portableand scalable. Moreover we can take advantage of new developments in VRML - such as proposedextensions to multi-user working that could allow collaboration between surgeon and trainee.

The major question to be resolved is the degree of realism that needs to be achieved. Perhaps too muchrealism can be a bad thing! In the area of rendering, there has been a powerful movement in favour ofnon-photorealistic rendering, where an expressionist style of rendering is used to create a moreinteresting final result. There may be a parallel in surgical simulation, where the key parameters in thesimulation are extracted and presented to the trainee in an ’expressive’ style. We would hope to discussthis at the workshop.

Acknowledgements

We are very grateful to the surgeons and radiologiss who have given generously of their time to help us- especially David Kessel (St James University Hospital) and Nick Phillips (Leeds General Infirmary).

June 1999

Visualisation in teaching-learning mathematics: the role of the computer

G. Chiappini and R.M. BottinoConsiglio Nazionale delle Ricerche

Istituto per la Matematica Applicata - Genova - Italy{chiappini, bottino}@ima.ge.cnr.it

Abstract

This paper is centred on the study of the role thatinformation visualisation can have in mathematicslearning. A theoretical analysis is worked out in orderto give account of the complex phenomena that takeplace in the classroom in order to allow students toconstruct a mental image of the mathematical objectinvolved in the activity. This theoretical analysis is thereference to study the different didactical functions thatinformation visualisation can develop in learningprocess. This different functions are discussed makingreference to the role that Cabri Gèométre can have inthe teaching/learning processes of Euclideangeometry.Keywords: Mathematics learning, Informationvisualisation, Visual imagery, mathematical discourse .

1. Introduction

We use the term visualisation to refer to the complexphenomena of visual imagery that plays a central rolein all meaning and understanding as well as in allreasoning. About the nature of images that the mindforms following an external stimulus, many studies,also on contrasting lines, were developed in the last 20years. Here we are interested in the phenomena ofvisual imagery that take place in maths learning and inparticular in the dialectic that develops betweendynamic external visual representations mediated bythe technology (information visualisation) and visualimagery.Speaking of external representations we mainly refer totwo or three-dimensional representations of someaspects of a mathematical structure. Suchrepresentations may be static or dynamic as in the caseof representations mediated by the computer. Dreyfusin [2] pointed out that the dialectic between externalvisual representation and visual imagery implies twomappings: from the mathematical structure to thevisual representation and from the visual representationto the mental image. While the first mapping can besubject to mathematical analysis (epistemologicalanalysis of the knowledge embedded in the structure ofthe visual representation), the second one is much moredifficult to analyse, since there is no direct access tomental images. So this latter mapping can only be

postulated on the basis of interpretations of how theexternal representations (graphics, signs, tables,drawings,…) are produced and used in thecommunicative context of the teaching/learningprocess and on the basis of the justifications thateveryone gives of this way of use.The two mappings evidenced by Dreyfus are importantwhile studying the conditions under which computer-supported visualisation is useful for mathematicslearning.This is true both for the design of new systems formathematics education and for the analysis of howsystems already available on the market can be used inthe classroom.In this work these two mappings will be discussed,analysing the roles assumed by informationvisualisation [7] in the teaching/learning processes ofmathematics. This analysis will be carried out on thebasis of the experience we have developed in thedesign, implementation and experimental evaluation ofcomputer-based visualisation systems for mathematicslearning [1].

2. Mathematical objects, visual representation,mental image

For the aims of our work we consider Mathematics as adomain of knowledge that is concerned with“mathematical objects”, that is to say with objects withcertain specific properties. Mathematical objects areabstract objects; indeed mathematics objects are notamenable to any concrete imagination or manipulation;they are immaterial, not tangible and accessible only toour thinkingIn mathematics learning, differently from the physicalconcrete world, the learning object cannot be shown inan obstensive way, can be only conjured up by meansof the use of external representations. There is not thepossibility of directly accessing that “thing” that wecan suppose to be the meaning of the representations.Mathematical concepts such as numbers, functions,vectors, (which are not objects in a usual manner, butwhich embody relationships) are not directly accessiblethrough everyday experience nor within intuitiveperception, as for instance real or physical objects are,but they have to be represented by signs or symbols.

This is true also in the case of Euclidean geometricallearning (as we will see in section 4), where theperception involved in managing externalrepresentations (drawings) can be also an obstacle forthe construction of a mental image, theoreticallyfounded, of the correspondent geometrical object(figure).Representations and symbols of mathematics establisha semiotic system which is of fundamental importancefor any mathematical activity.According to this epistemological positionmathematical knowledge is not simply a ready madeproduct that can be directly introduced into processesof teaching and learning. The new mathematicalknowledge will only be actively constructed, in socialinteraction, by the student in his or her learning processwithin an activity.For example, in the approach to rational numbers therelationship that takes place between representationand the mathematical object which it refers to (rationalnumbers), is very complex. The discourse aboutrational numbers suggests that these abstract objectsare unique entities; on the other hand we have variousrepresentations for the rational numbers: the commonfraction symbol, points on the number line,materialisation of different kinds, classes of pairs ofintegers; these material objects of representations donot enjoy all the properties we attribute to rationalnumbers [3].It is not easy for students to understand that 2/3 and 4/6are two different fractions which are quivalent sincethey refer to the same object, that is to say to the samerational number. Showing that 2/3 is obtained from 4/6dividing the numerator and the denominator by 2 givesus a method that permits us to justify the conditionsunder which the two fractions are equivalent but wecannot work out an obstensive test of such an objectand of its properties. We do not have imaginativeaccess to anything which we could consider an imageof the rational number represented by 2/3 or 4/6.The image of this mathematical object emerges in thedialectic between understanding and expressionrelating to the mathematical activity in which theparticipants are involved. This image entails anorientation to negotiations with oneself about meaning,something that is outside the experience of schoolstudents [6].The image of the mathematical object is strictly linkedto the image of the mathematical activity that isnegotiated by the participants in school practice. It isconditioned by social interaction; on one hand thestructure of the external representations providemeaning for the mathematical discourse about themathematical object involved in the activity and on theother hand the mathematical discourse contributes tostructure an image of the mathematical activity, that is

to say it contributes to structure an image of the objectof this activity, that is a mathematical object. It isimportant to observe that mathematical discourse thatemerges in social interaction within the activity isqualitatively different from any representation.Understanding what a mathematical object is dependscrucially on a very specific way of viewing andtreating representations and their related mental imageswithin a mathematical discourse.In other words, taking into account school practice inwhich students and teacher are involved, therelationships between external visual representation,mathematical discourse and mental images can besketched in this way.Within an activity, the structure of externalrepresentations mediate the possibility to develop ameaningful mathematical discourse about theproperties of the mathematical object starting fromhow the properties of the external representation areused in practice.Mathematical discourse (which is always mediated bythe reference to the mathematical object) allows ametaphoric use of the concrete meaning developed byworking with external representations within thestructure of an activity. It allows us to associateinterpretations to the external representation used in theactivity which can be justified on the basis of actionsand goals as actually generated within an activity (inrelation to some problem or task) and on the basis ofcultural and historical considerations on the value andproperties of the mathematical object involved in theactivity. In this way the mathematical discoursecontributes to transform an external representation intoa mental image of a mathematical object, that is to sayinto an image of the abstract object which transcendsthe structure and the characteristics of the externalrepresentation.From a psychological point of view the dialecticbetween external representation and mathematicaldiscourse mediates the possibility for the subject tocontrol, on the basis of external stimuli (externalrepresentations), the accordance of the mental image ofthe mathematical object involved in the activity withthe shared ideas of the society.Hence according to [3] grasping the properties of amathematical object is always the result of interplaybetween visual and diagrammatic aspects involved inmanaging external representations and prepositionalaspects of the discourse with respect to themathematical object involved in the activity.

3. Information visualisation and mathematicslearning

The design or the use of a computer-based system formathematics learning requires careful consideration of

the conditions under which the characteristics of formand interactivity of a system can develop, within anactivity, a dynamic relationship between externalrepresentation and mental image which is effective forlearning.According to [5] we use the term informationvisualisation to intend the use of computer supported,interactive, visual representation of abstract data toamplify cognition. This definition is particularlyappropriate when it refers to the teaching and learningprocesses of mathematics. As previously pointed out,mathematical objects are not amenable to any visualperception or manipulation. The traditional approach tomathematics knowledge is a symbolic re-constructiveapproach and it is developed inside the interactionbetween the student and the teacher, usually accordingto a transmissive teaching strategy. In this approachstudents have few opportunities of exploring thefunctionality of the symbolic representation at handand of reflecting on the properties and characteristicsof its structure. The prescriptive character of themathematical discourse that emerges in the classroomfocuses the attention on the system of rules attached tothe symbolic representation at hand and leaves in thebackground the construction of a mental image of themathematical object involved in the activity.Information visualisation can allow the student toaccess mathematical knowledge integrating thesymbolic re-constructive approach with a motor-perceptive one. This latter approach involves actionsand perceptions and produces learning based on doing,touching, moving and seeing. As already pointed outby Kaput [4[, a visual representation system based ondirect manipulation interface is an interactive mediumwhich responds to the user's action and which offersthe possibility to create new notational systems or tointroduce a new dimension, movement, withintraditional ones.Information visualisation offers the concrete possibilityto implement better and more easily classical visualrepresentation mathematics formats (graphs, drawings,tables, etc.) but also to enrich them with additionalfeatures such as movement and interactivity and tointegrate in the same environment (or in interconnectedenvironments) multiple formats of representations.Within the context of use of the system, informationvisualisation can support different didactical functionswhich are crucial in the teaching and learningprocesses of mathematics.Information visualisation can offer ways which allowstudents to explore the knowledge domain, embeddedin the system (exploratory function). Systems thatpresent these features have been demonstrated to bevery effective for learning mathematics: they aredefined as microworld based systems. The pedagogicalobjective of microworlds is to offer students a space in

which they can use visualisation supported by thecomputer in order to explore and manage freely anenvironment designed to address the construction ofsome mathematical knowledge. Other systems thatpresent these features are systems for simulation.Another important didactical function of informationvisualisation is to offer expressive ways to allowstudents to externalise their own knowledge of adomain (expressive function). This function is presentwhen representative tools, which student can easilycontrol both on an operative and a conceptual level, aremade available within a system.A third didactical function regards the possibility tooffer ways to validate the solution strategy involved inthe problem at hand (validation function). In this caseinformation visualisation makes available verificationmethods that are checked on a perceptive level and areable to test the strategy developed.Systems that present this didactical function makeavailable tools that allow students to justify theirsolution procedures on the basis of verificationmethods that are internal to the system.A fourth didactical function is connected to thepossibility to have available tools that allow the studentto store the solution steps performed in order tosuccessively re-visualise them in sequence. In this wayit is possible to re-construct the story of the solutionprocess performed. Systems which make available thiskind of tools make possible to objectify, de-personaliseand de-contextualize a solution procedure in order tosuccessively transform it into an object of discussion inthe social context of the class (supporting mathematicaldiscourse function). This discussion can have differentaims such as the comparison of strategies, the analysisof the properties (and of the their relationships)involved in the procedures, the re-interpretation of suchproperties in relation to a theory of reference.According to the nature of the knowledge domaininvolved in learning, the different described didacticalfunctions of information visualisation can be pursuedin different ways. In the following, we will try to betterexplicit how information visualisation can affectlearning mathematics, analysing these differentfunctions in a specific domain of knowledge supportedby computer.

4. Information visualisation and learningEuclidean geometry

Geometry is a knowledge domain in whichvisualisation within a microworlds has brought toimpressive progress with the development of theconcept of “dynamic geometry” exemplified bysystems such, for example, Cabri Gèométre. Theresearches worked out in relation to the use of thissystem put in evidence that its mediation can modify

the way to enter into contact with geometricalknowledge. The changes brought into the didacticalactivity by the use of Cabri can be debated in terms of:* New status assumed by the geometrical constructionas mediated by the system;* New possibilities of development of a mental imageof a geometrical construction, theorem and proof, asmediated by the system.In this framework we note that the first aspect can beput into relationship with the mapping (as reported insection 1) from the mathematical structure to the visualrepresentation while the second aspect concerns themapping from the visual representation to the mentalimage of geometrical construction, theorem and proof.The relationship between geometrical construction andvisual representation available with Cabri isparticularly interesting.By means of Cabri it is possible to perform every typesof geometrical construction, within Euclideangeometry. Performing a geometrical construction bymeans of Cabri does not implicate aspects of measure,but it is strictly connected with the deep structure ofEuclidean geometry. The fundamental element ofevery construction is the point. Some objects aredirectly defined in terms of points (i.e. a straight line isdefined by means of two points). Other objects aredefined as functions of some other objects alreadyconstructed (i.e. a straight line passing for a pointperpendicular to a given line) [7].Cabri makes available primitives deeply linked withthe axioms of the Euclidean geometry but also a newvisual direct manipulation opportunity: to drag thevariable elements of the geometrical construction onthe screen. In this way students can observe whichproperties are preserved when the construction ismodified with the drag action. The movementproduced by the drag action is a way to externalise theset of the relationships that define a figure.These representation tools of Cabri can be used in bothexploratory and expressive modes.In the exploratory modes, the use of theserepresentation tools containing domain models permitthe student to examine the consequences of differentmodel, some of which fit and other of which conflictwith the student own ideas [9].In the expressive mode student can examine their ownknowledge of geometrical construction by beingencouraged to experiment with their own theorizing[9].Mariotti [8] highlights that the drag action of Cabri hasa crucial importance both from a didactical and anepistemological point of view. By means of this actionstudents can validate the constructions performed; theconstruction task is solved when the drawing getthrough the drag test, that is when the properties of the

figure, under the drag action, are preserved (validationfunction).Moreover the use of this new dimension, themovement, in the body of Euclidean geometry, makesavailable new visual tools to overcome epistemologicalobstacles connected with the relationship betweendrawing and figure in the domain of Euclideangeometry.As stressed by Laborde [7], the drawing refers to amaterial entity with which one has a perceptiverelation, figures refer to theoretical objects, that is toabstract geometrical object that are described by textswhich define them.In traditional didactical practice we note that severaldifficulties emerge in the geometrical constructiontasks. Students work on the material drawing while theteacher expects they work on the figure or on thedescription of the figure. In other cases students do notconsider that a geometrical construction task caninvolve the use of geometrical properties; rather theyintend it as the request to produce a material visuallycorrect drawing.According to this framework, the opportunity ofdragging the variable elements of a construction is animportant tool for favouring the evolution of students'mental images, which are related, with the notion ofgeometrical figure and with the task itself ofgeometrical construction (exploratory and expressivefunction)As an example the way in which student’s approachgeometrical constructions with Cabri can beconsidered. At the beginning students work on thedrawing on the basis of perceptual and figural stimulireceived from the screen instead of to perform aconceptual control of the perception, exploiting theaction possibilities offered by the system. Only whenthey realise that their empirical strategy does not getthrough the dragging test, students, in general, begin toapproach the task by using constructive methods whichallow to maintain the properties of the figure when thedrawing is dragged on the screen. It is in this momentthat the relationship between figure and drawing isgoing to be modified and the student elaborates a newmental image of geometrical constructions.The evolution of the mental image of geometricalconstructions is crucial in geometry. As the matter offact, each geometrical construction incorporates atheoretical meaning, which goes beyond the practicaltask of its construction. Mariotti in [8] shows that thereis always a correspondence between the specific toolsand rules used for the construction and a set of axiomswhich structure a part of a theory. Inside this theory avalid construction can always be put in correspondencewith a theoremGeometrical learning needs to move from context-based methods of validation (e.g. the dragging action

of Cabri) to theory-based justifications (deductivemethod). This passage is neither simple nor automatic.It requires an educational approach able to support theacquisition of the deductive method. In this frameworkit is important specify the role that Cabri can assume insupporting this didactical approach.In Cabri, for example, an important command isavailable at this regard: the "History command". Thiscommand permits to visualise step by step on thescreen all the actions performed by the user to obtain agiven geometrical construction. Such a command candevelop some important function from a didacticalpoint of view.As put in evidence by Mariotti [8], the Historycommand allows to develop interpretations andanticipation acts which can favour the passage fromjustifications based on physical actions and graphicaleffects to justifications based on a deductive model. Inthis passage there is the development of the capabilityto justify the validity of a procedure at the conceptuallevel (demonstration) on the basis of initial hypothesis,geometrical axioms and previously demonstratedtheorems.This passage cannot be spontaneously built as theresult of the student-software interaction (e.g. as theresult of having observed on the screen constructionsthat can be modified preserving some properties). Thedeductive method is built in the social interaction whenstudents can experiment, under the teacher's guidance,the importance of justifying the accomplishedprocedure according to methods, which go beyondperceptual validation.In the Cabri environment the History command is afundamental tool at this regard. With this command theaccomplished procedure becomes objective, de-personalised and de-contextualised.In this way it is possible to focalise the attention on theproperties of the construction and on their relationshipsand to consider them in the light of a geometricaltheory and of previously demonstrated facts.The History command is an important tool to maintainthe mathematical discourse, which is developed in theclass on the value, methods and techniques ofgeometrical demonstration integrated and "in focus"(supporting mathematical discourse function). This canbe obtained thanks to the possibility offered ofanalysing and putting in relation the accomplishedconstructions

References

[1] Bottino R.M., Chiappini G., "ARI-LAB:models issues and strategies in the design ofa multiple-tools problem solvingenvironment", Instructional Science, Vol.

23, n°1-3, 1995, Kluwer AcademicPublishers, pp. 7-23

[2] Dreyfus T., Immagery for Diagrams in R.Sutherland and J. Mason (eds) Exploitingmental imagery with computer inMathematics education, NATO ASI Series,Springer - Verlag, Berlin, 1995, Series F,Vol 138, 3-19

[3] Dorfler W., Mathematical objects,representations, and imagery in R.Sutherland and J. Mason (eds) Exploitingmental imagery with computer inMathematics education, Nato Asi Series F,Vol. 138, Berlin: Springer Verlag, 82-94,1995.

[4] Kaput J.J, Overcoming phisicality and theexternal present: cybernetic manipulatives,in R. Sutherland & J. Mason (eds):Exploiting mental Imagery with Computersin Mathematics Education, Nato Asi SeriesF, Vol. 138, Berlin: Springer Verlag, 161-177, 1995

[5] Card S. K., Mackinlay J. D., ShneidermanB., Information Visualisation, MorganKaufmann Publisher, 1999

[6] Thompson P. W., Imagery and thedevelopment of mathematical reasoning, inSteff, Nesher, Cobb, Goldin, Greer (eds)Theories of mathematical learning, LEAPublisher, 267-285, 1996

[7] Laborde C., The computer as part of thelearning environment: the case of geometry,in Keitel C. & Ruthven K. (eds.), Learningfrom Computers: Mathematics Educationand Technology, Nato Asi Series F, Vol.121, Berlin: Springer Verlag, 48-67, 1993

[8] Mariotti M.A., Introduzione alladimostrazione all’inizio della scuolasecondaria superiore, L’Insegnamento dellaMatematica e delle Scienze Integrate, Vol.21B, N 3, pp 209-252, 1998

[9] Bliss J. , Externalizing thinking thoughmodelling: ESRC tools for exploratorylearning research program, in Vosniadou, Decorte, Glaser, Mandl (eds), Internationalperspectiveson the design of tecnology-supported learning enviroments, LEAPublisher, 25-40, 1996

Virtual Environment of Water Moleculesfor

Learning and Teaching Science

Jorge F. TrindadeInstituto Politécnico da Guarda

Escola Superior de Tecnologia e GestãoGuarda, Portugal.

jtrindade@ipg.pt

Carlos FiolhaisDepartamento de Física

Universidade de CoimbraCoimbra, Portugal.tcarlos@hydra.ci.uc.pt

Victor GilDepartamento de QuímicaUniversidade de Coimbra

Coimbra, Portugal.vgil@cygnus.ci.uc.pt

José C. TeixeiraCentro de Computação Gráfica

Coimbra, Portugal.teixeira@ccg.uc.pt

Abstract

Virtual reality is an emerging computer visualizationtechnology which allows educators to place theirstudents into instructional environments heretoforedifficult or impossible to achieve. In order to take fulladvantage of this technology, a virtual environment,"Virtual Water", is being developed. Virtual reality hasbeen recently employed in a few educationalapplications, but the project presented here is the firstapplication we know of combining aspects as atomicand molecular orbits, electron densities, watermolecular dynamics for the liquid and gaseous phasesand phase transitions.Keywords: virtual reality, water, molecular dynamics,quantum mechanics.

Introduction

In the past fifteen years, research on physics educationhas brought a lot of information about the difficultiesstudents have in learning physics. At the same time, theongoing revolution in information technology has ledto new tools for creating innovative educationalenvironments. In response to these developments, awide variety of new models of physics instruction areavailable.

People's understanding of what computers can do hasshifted dramatically as the size and cost of thesedevices has decreased while their power has increased.At the beginning, computers were seen as number-crunching machines, but now we have tools tomanipulate information, namely in the graphic form.Up to now, the use of computational means in Physicseducation stood mainly on the creation of 2Drepresentations that the students could use to buildmore refined mental models. However, recent advanceshave created new possibilities and the visualization of3D objects and data becomes increasingly important in

learning several scientific subjects (in particular,atomic and molecular science, fluid dynamics, etc.).With virtual reality (VR), the visualization of complexdata and the building of more adequate conceptualmodels is possible [1, 2].

VR is one of the most significative steps towards anatural human-computer communication, allowing aneasy presentation and an intuitive grasp of complexdata. Virtual environments can represent variousaspects of a natural environment or even a totallyartificial world. The inclusion of haptic informationand direct manipulation contributes to the impressionof being immersed in a real situation [3]. Perception, aslearning, is an individual act. Since each personevolves a unique psychomotor approach to interactingwith a physical environment, individuals have morevaried responses to 3D interfaces than to the standard2D graphical interfaces with menus, windows, andmouse.

The Physics Department of the University of Coimbra,Portugal, the Exploratory Henry the Navigator, theComputer Graphics Center (both also in Coimbra) andthe High Education School for Technology andManagement of Guarda, are developing the "VirtualWater" project, a virtual environment devoted to thelearning of Physics and Chemistry. This work involvesthe know-how of different fields (computationalsimulation of physical systems, computer graphics andscience education) in order to arrive at a usefulscientific-pedagogic visualization of water.

The main goal is to enable the user to visit virtualattractive and enjoyable sceneries where he can learnthe constitution and properties of water. The user willbe able to interact and “change” the environment tobetter understand the structure of water. The virtualenvironment is realistic and comprehensive, allowingfor exploring the atomic constituents of water as well

as macroscopic structures. With this environment, thestudy of the atomic units, the molecule geometry, themolecular orbitals, the electron density, and molecularnormal modes is facilitated. Using simplifiedequations, as those of Newtonian mechanics with theLennard-Jones potential, simulation of the moleculardynamics of the solid, liquid and gaseous phases andphase transitions, is also available. Realisticsimulations use more complicated forces [4].

Description of the Project

The study of water is motivated by the fact that a betterunderstanding of it, in its different functions andaggregation states, is only possible if the watermolecule itself and the behavior of a set of moleculesin given conditions are understood. With the speed ofmodern computing resources, it is becoming commonin research to model aqueous systems atom by atom,moving each molecule in response to the forces actingon it. These advances should be followed by educationapplications.

2.1. Our ProblemInnovative practices using computers are being triedfor teaching and learning science [5]. VR has beentouted as a powerful teaching and training tool because([6], [7], [8]):• Supports direct experience of phenomena.• Is 3D.• Allow for multiple frames of reference.• Offers multisensory communication.• Is physical immersive.

However, current immersive VR technology hasseveral limitations that may impede usability andlearning ([9]):• Head-mounted displays and gesture devices

interfere with interaction.• Input devices have restricted fidelity and

versatility.• The display capabilities are somewhat limited.• Multisensory inputs can interact to cause

unintended sensations and perceptions.

Through the design of "Virtual Water" we have startedto determine how the characteristics of VR influenceusability and learning. At present, achieving VR'spotential to enhance learning requires transcending theinterface barriers through paying careful attention tousability issues.

2.2. Our ApproachTo design a educational virtual world, we have adoptedan iterative learner-centered design approach thatfocuses on usability and learning. The main goals inproducing our application are:• To provide an educational environment for

students to explore some microscopic and abstractconcepts, which they are teached in class but arefar away from daily experience.

• To develop a practical knowledge concerning theapplication of VR techniques in education.

• To contribute with data on the pedagogicalusefulness of VR. People in the field have anintuition that VR can have a strong impact onlearning. But believing in VR is not enough: itsusefulness has to be proved.

2.3. System Configuration and Models DevelopmentThe interface of our immersive, multisensoryenvironment is typical of current high-end VR."Virtual Water's" hardware architecture includes aPentium II processor with 3D graphics accelerator. Forthe navigation and immersion in the virtualenvironment, we use the Head Mounted Display V6from Virtual Research, as well as one Cyberglove withcybertouch (for haptic information), from VirtualTechnologies, and a Polhemus Isotrack II magneticorientation position sensor for two receptors (seeFigure 1). This class of hardware will not be commonin classes in the next years, but it allows us to deliver aproduct with minimal quality.

Figure 1. The virtual reality hardware interfaces.

Concerning the software, we use WorldToolkit (fromSense8), that serves the definition and creation of thevirtual scenarios, and other packages for modeldevelopment. For the design of the first part (quantummechanics) of "Virtual Water", we use mainly thefreeware PC Gamess [10], that performs the watermolecule calculations (namely geometry optimization,

electron density, etc.) and the freeware Molden [11]package for molecular representation. For the secondpart (molecular dynamics), we use commercialsoftware packages (Mathcad, 3D Studio Max andAutocad) for development of models and optimizationand Visual C++ for implementing the moleculardynamics algorithm.

2.4. The Training EnvironmentScenery exploration is preceded by navigation in atraining environment. The first "room" which usersencounter is the training environment, which containsknown objects. The goal is to overcome thedisorientation problems which some users experiencewhen first using VR. This virtual scenery is designed tobe a simple, familiar environment in which users canbecome comfortable with the VR hardware beforemoving on to more complex an abstract environments.

2.5. The Quantum Mechanics Virtual EnvironmentIn this scenery we deal with the geometry of the watermolecule (see Figure 2-a), its electron density (seeFigure 2-b), and molecular orbitals (see Figure 2-c andd). The 3D models were produced with PC Gamess (aprogram that performs ab initio calculations to obtainthe molecule geometry), and Molden (a program thatenables the visualization).

In this virtual environment students can build andtravel through molecular orbitals, ad molecular densityand better understand the water molecular structure inorder to clarify the chemical bonding.

a)

b)

c)

d)

Figure 2. Some 3D models of the quantummechanics virtual environment: a) ball-and-stickrepresentation of water molecule; b) equidensitysurface representation of the same molecule; c) thethird and fourth occupied molecular orbitalrepresentation of water molecule (the total electrondensity is obtained adding up the occupiedmolecular orbits).

2.6. The Molecular Dynamics Virtual EnvironmentWe aim at understanding some water properties bydirect simulation. We assume that the moleculardynamics can be treated classically because realisticsimulations (with quantum effects) use complicatedforces [4] and are much more computationaldemanding. We use Newton's equations of motion

��−=j

ijii ru

dt

rdm )(

2

2 rr

with the Lennard-Jones potential,

√↵��

� −= 612 )()(4)(

rrru

σσε

(the ε and σ are constants).

The repulsion at small distances is a consequence ofthe Pauli exclusion principle. The weak attraction atlarger distances is due to the mutual polarization ofeach molecule.

Newton's equations are solved for each moleculestarting from initial positions and velocities and usingthe force acting on each molecule

rrrr

ru)r

���

�

���

�√↵��

�−√

↵��

�=�−

612

224

)(σσε .

Then, we calculate the positions

2, )()()(2)( t

m

tFttxtxttx ix

iii ∆+∆−−=∆+

and velocities

t

ttxttxtv ii

ix ∆∆−−∆+=

2

)()()(,

of each molecule at successive times.

The user is able to interact and change the environmentin order to study the liquid and gaseous phases andphases transitions (see Figure 3-a). The solid phase isalso examined (see Figure 3-b).

One of the new aspects of this pedagogical work is the3D representation of the ensemble of molecules insteadof the usual 2D. The same algorithm has been appliedto the liquid and gas phases, emphasizing theirsimilarity.

In order to assure a real-time rendering (the level ofdetail which is attainable at a practical speed), thenumber of molecules in each phase or phase transitionsimulation has been carefully tested. For example, forthe gaseous phase simulations we have used 20molecules. This restriction is the compromise betweenthe computer graphics capability, molecular modelsand real-time calculations. Contrary to most computerapplications, VR must recalculate the user's view foreach frame update, taking account considerations aslight sources, shadings, distance from the user, etc. Tobe effective, all this must be performed several timesper second in addition to any calculations that have tobe made.

a)

b)

Figure 3. Images from the molecular dynamicsenvironment: a) the gaseous phase, with the balland stick model, showing twenty molecules in abox; b) The ice phase, with the same number ofmolecules, but with the electron densityrepresentation. These pictures were created usingthe same software as in Figure 1, being thedynamics implemented in Visual C++.

Conclusion

The use of 3D graphics seems to be a powerful tool forvisualizing and understanding complex and/or abstractinformation. Immersion is a recent aspect which has tobe better explored and evaluated. A virtualenvironment for teaching and learning Physics andChemistry is being developed to study the possibilitiesof VR in education. The main objectives are:• Get new means for teaching and learning the

physics and chemistry of water.• To apply VR as an educational tool, in order to

determine which aspects of VR provide the mosteffective educational benefits, and to learn thestrengths and weaknesses of this technology in aneducational setting.

• To build knowledge of VR techniques and toolswhich can later be applied to other problems.

Feedback from students still needs to be collected andanalyzed.

Acknowledgements

We wish to acknowledge the assistance of the studentAndré Dias, who has a developed the moleculardynamics part of the "Virtual Water".

References

[1] J. Trindade and C. Fiolhais. A realidadevirtual no ensino e na aprendizagem daFísica e da Química. Gazeta da Física 19,pages 11-15, 1996.

[2] C. Fiolhais and J. Trindade. Use ofComputers in Physics education.Proceedings of the "Euroconference'98 –New Technologies for Higher Education".Univ. Aveiro: ed. A. Ferrari, Aveiro, 1999.

[3] J. C. Teixeira, A. Côrte-Real, P. Bernardesand F. Macedo. Santa Clara-a-Velha: virtualenvironments and cultural heritage.Proceedings of CAD & Graphics'97, FifthInternational Conference on CAD & CG,pages 81-88, Shenzhen, China, December,1997.

[4] M. Laasonen, M. Sprik and M. Parrinelo."Ab initio" liquid water. J. Chem. Phys. 99(11), pages 9080-9089, 1993.

[5] J. C. Teixeira. Environments for teachingcomputer graphics: an experience. Comput.& Graphics 20 (6), pages 927-935, 1996.

[6] M. Bricken and C. Byrne. Summer studentsin virtual reality. In Wexelblat, A. (Ed.),Virtual Reality: Applications and

Exploration, pages 199-218, New York:Academic Press, 1993.

[7] T. Erickson. Artificial realities as datavisualization environments. In Wexelblat, A.(Ed.), Virtual Reality: Applications andExploration, pages 1-22, New York:Academic Press, 1993.

[8] R. Kalawsky. The Science of Virtual Realityand Virtual Environments. A Technical,Scientific and Engineering Reference onVirtual Environments. Addison-WesleyPublishing Company, 1993.

[9] C. Wickens. Virtual Reality and Education.Proceedings of the IEEE InternationalConference on Systems, Man, andCybernetics, 1, pages 842-847, New York:IEEE Press, 1992.

[10] PC Gamess, a program for ab initio quantumchemistry, written by Alex. A. Granovski,Moscow State University.

[11] Molden, a package for displayingMOLecular DENsity, written by G.Schaftenaar, CAOS/CAM Center Nijmegen,Toernooiveld, Nijmegen, The Netherlands.

Evaluation of Driving Education Methods in a Driving Simulator

J. Miguel Leitão1,3, Alexandra Moreira2, Jorge A. Santos 2, A. Augusto Sousa3,4 and F. Nunes Ferreira4

1 Departamento de Engenharia Electrotécnica, Instituto Superior de Engenharia do Porto, Portugal2 Instituto de Educação e Psicologia da Universidade do Minho, Braga, Portugal

3 Instituto de Engenharia de Sistemas e Computadores do Porto, Portugal4 Faculdade de Engenharia da Universidade do Porto, Portugal

jml@ipp.pt

Abstract

This paper presents the development of a realisticdriving simulator and its use in the study of drivingeducation methods. In this study, a large set of personswithout driving license will be used. Each person willbe submitted to a driving course that may includedriving lessons in a real car and / or in the drivingsimulator. The realism used in the simulation can beadjusted between the classical non-interactive, filmbased training and the full realism achieved by ourdriving simulator. Each driving education method willbe evaluated by comparing driver’s competence beforeand after the applied course.

We expect to determine the influence of the simulationrealism in the learning process, and to understand howand when the driving simulator should be used forbetter performance in the education process.

1. Introduction

The large number of accidents and dead occurred dueto traffic circulation in public ways is one of the mostconcerning problems of today societies. It has beencommonly accepted that one method to reduce theselosses is the improvement in driver’s education andselection. To do so, and due to recent advances andspread of image generation systems, interactive drivingsimulators are being more and more used in drivingeducation tasks.

It is commonly accepted that the use of drivingsimulators presents some advantages over thetraditional methods of drive learning. Because of theirvirtual nature, the risk of damage due to incompetentdriving is null. This usually allows a reduction in thecost of driving lessons or a longer and richer trainingbut imposes a very hard limit to the cost of a drivingsimulator. According to professionals in the area ofdriving education, to explore its advantages in drivingeducation use, a driving simulator must not be muchmore expensive than real vehicle.

Although the recent advances in computing and imagegeneration systems, to keep the cost of a drivingsimulator compatible with the education uses, manysimplifications must be done. For example, reductionsin field of view, image resolution, vehicle's dynamicbehavior and scenario reactivity are common in todayeducational driving simulators.

This paper presents the development of a realisticdriving simulator and its use in the study of drivingeducation methods. This study will be done using dataacquired from a large set of persons without drivinglicense. Each person will be submitted to a drivingcourse that may include driving lessons in a real carand / or in the driving simulator. The realism used inthe simulation will be adjusted between the classicalnon-interactive, film based training and the full realismthat can be achieved by our driving simulator. Eachdriving education method will be evaluated bycomparing driver’s competence before and after theapplied course.

As results of this study, we expect to determine theinfluence of the simulation realism over the learningprocess, and to understand how and when the drivingsimulator should be used to maximize the drivingeducation’s productivity.

The state of the art in the field is summarized inSection 2. Section 3 discusses the use of drivingsimulators in driving training and driver’s evaluation.Section 4 presents a short description of DriS, its goalsand applications. The study, being prepared, aboutdriving education methods and its evaluation isdetailed in section 5.

2. Related work

Driving simulators are hi-complex real-time, imersivesystems [1][10]. The improvement in realism and thevalidation of their performance are the major concernsin development of any scientific driving simulator[3][4][12].

The usability of simulators in training and in evaluationof human performance is already studied and generallyaccepted under several restrictions [14][9]. In drivingeducation, it is today common to find the use of drivingsimulators either with training or with evaluationproposes. Some experiences are known that analyzediscrepancies between human driving in real and insimulated environment [11][13], but no studies areknow that evaluate how these discrepancies affect thedriving education process.

3. Driving Simulators

In the last years, we assisted a tremendousdevelopment in computing and image generationtechnology that allowed the proliferation of VRsystems with acceptable quality for some educationpurposes. Even thought, the development andmaintenance of a realistic driving simulator is still avery expensive task. So, to keep the cost of a drivingsimulator compatible with the education uses, manysimplifications must be done.

For example, most driving simulators used for drivereducation present only a narrow filed of view. Thisreduction clearly degrades the perception of thesurrounding environment by the human driver and alsolimits the amount of information that he must receiveand process. Also, the generation of the backwardimages presented in mirrors is crucial, for instance, fortraining overtaking maneuvers.

The image resolution used in today driving simulatorsis always poorer than the one obtained by the directobservation of the real world. This results in a differentdriver behavior due to the lack of obstacle perception.It is known that, when approaching an obstacle, adriver breaks sooner in a driving simulator than in areal car [13].

Reductions in vehicle's dynamic behavior and scenarioreactivity are also common in today educationaldriving simulators.

All these, and other simplifications affect human driverreactions but, it is accepted that they do not invalidatethe data we can get from the simulated driving task andthe knowledge acquired from training. It is known that,when correctly used, even a simplified drivingsimulator can be useful either in driver performanceanalyses as in driving education. A big challenge that isproposed to every driving educator is the choice ofhow and when to use driving simulator in order tomaximize the education productivity. This choiceimplies a selection of simulation realism and its tradeoff with the simulator cost. Then a selection of trainingexercises must be made but this selection is usually

limited by the realism achieved by the possiblyavailable driving simulator. Because a simulatedtraining is still far from completely substitute a realexperience, decisions must also be taken on how thesetwo kind training should interact.

4. DriS - Driving Simulator

A description of our driving simulator (DriS) can befound in [2]. Its main core runs on a SGI Onyx RealityEngine 2 graphical workstation (Figure 1). Thisworkstation holds the scene database, and performs thesimulation and the image synthesis tasks. It isconnected to the sound PC and to the cockpit PC bylocal area network. Communications are implementedsynchronously with the rendering process, using asocket library.

Figure 1: DriS driving simulator.

In DriS, the driver sees a realistic image either in alarge screen display or in a HMD. The generated imageis applied to a large flat screen (3.40mx2.00m) by avideo projector. Typical experiments are usuallyperformed with resolution of 1280x1024, with animage update of 20 images per second, and a framerate of 60 frames per second [15].

High quality audio information is also supplied ineither stereo sound with headphones or in multi-channel surround sound using loudspeakers, and anadditional channel for the typical sound of the vehicle[6].

DriS is being developed by a large team of researchersfrom several institutions and different scientific areas.The development team includes experts in computersystems, image analysis, computer graphics, road andtraffic engineering and psychology. Their main goalsare the study of driver behavior and road analysesunder conditions that are difficult or even impossible toreproduce:• risk situations

• accident• new roads• new managing systems• new road signals• new vehicle models

4.1. Applications of DriS

In [7], Noriega and others presented a study ofvehicle's motion detection with concurrent self motionwith three kinds of road pavement:• concrete pavement• bituminous pavement• bituminous pavement with chromatic bands

The DriS driving simulator allowed an easypreparation of the stimulus presented to each one of the106 persons with driving license that participated inthis study. It was concluded that the pavement withchromatic bands produces a higher number of wrongdetections of vehicle's motion, while differences inimage contrast were considered not relevant.

A new study is now running to evaluate the influenceof external publicity panels (outdoors) in drivingperformance and several others are being prepared [5].One of most important studies is the evaluation ofdriving education methods that is detailed in this paper.

5. Evaluation of Driving Education Methods

This section presents the study of driving educationmethods that is now being prepared. This study will bedone using data acquired from a large set of personswithout driving license. Each person will be submittedto a driving course that may include driving lessons ina real car and / or in the driving simulator. The realismused in the simulation will be adjusted between theclassical non-interactive, film based training and thefull realism that can be achieved by our drivingsimulator. Each driving education method will beevaluated by comparing driver’s competence beforeand after the applied course.

5.1. Driving Environment

A driving simulator will be used (DriS), and the virtualenvironment will try to reproduce a typical street fromPorto. For this propose, we choused some well-knownstreets and avenues from Porto located in the Foz doDouro area. This area is traditionally one of the mostused areas in Porto for driving lesson purposes.

Roads were modeled using real field dimensions andbuildings were finished using photographic texturestaken from some choused streets:• Avenida do Brasil

• Avenida de Montevideu• Passeio Alegre.

Figure 2 presents a view of a modeled road. Severalautonomous vehicles were added to compose theenvironment traffic. These vehicles are driven byvirtual drivers that try to emulate human drivers [8].

Figure 2. View of a modeled road.

5.2. Methodology

This study will be done using data acquired from alarge set of persons (subjects) without driving license.Each person will be submitted to a driving course thatmay include driving lessons in a real car and / or in thedriving simulator. The realism used in the simulationwill be adjusted between the classical non-interactive,film based training and the full realism that can beachieved by our driving simulator. Each drivingeducation method will be evaluated by comparingdriver’s competence before and after the appliedcourse.

The subjects set is being formed using people that meetall the following requirements:• have not driving license• want to have driving license• are, preferable, between 16 and 22 years old• have not visual or physics deficiencies

The study will be performed in two identicalexperiments. The preliminary experiment will be usedto select the course curriculums that will be used in thefinal experiment. The preliminary experiment is nowstarting and will take about 9 months long.

Each subject will first be evaluated in drivingperformance terms and assigned randomly into one of4 groups (Figure 3). Some effort must be done in orderto achieve groups with the same amount of subjectsand with similar driving and intellectual skills. We plan

to use near 20 persons in the preliminary experimentand about 200 in the final one.

Then a different driving course will be applied to eachsubject group.

Course 1 will be a traditional driving course,performed at several driving schools and with differentdriving teachers. In Portugal the minimum curriculumof a driving course includes 25 hours practical lessonsbut it is known that this minimum is sometimes notneeded to achieve the required driving competence. Inour study, the final evaluation will take place after 15hours of practical driving lessons.

Course 4Group4

Group1

Course 3Group3

Course 2

Course 1

Eval

uati

on

Eval

uati

on

Gro

up fo

rm

atio

n

Subjectsset

Group2

Figure 3: Experiment workflow.

Course 2 will consist of 30 hours of driving lessonsusing an interactive driving simulator. Lessons will besupervised by a driving teacher and will includesimulated driving practice under several conditions of:• traffic intensity,• traffic velocity,• own vehicle characteristics

Course 3 will consist of 40 hours of driving lessonsusing a non-interactive driving simulator. Lessons willbe supervised by a driving teacher and will includesimulated driving practice under several conditions of:• traffic intensity,• traffic velocity,• own vehicle characteristics

5.3. Evaluations

Initial and final driving performance evaluations willbe done using data acquired from the simulated drivingin a realistic driving simulator.

Each subject will be asked to follow a previous route,in order to adapt to the simulator. Then, in theexperimental route, the driver will follow a first roadwithout other vehicles, a second one with low trafficintensity, and a third with high traffic intensity.

During the experimental route, several drivingparameters will be measured and stored for furtheranalyses.

Measures can be divided into three groups: measuresreported by the driver, a rating scale of mental effort(RSME); physiological measures, eye movements(with a NAC eye tracking recorder); and performancemeasures.

The rating scale of mental effort will be monitoredfrom several electro-magnetic sensors located in thesubject's head, using dedicated specialized equipment(Figure 4) that produces a graphical output printed inpaper.

Electromagnetic sensors

Data acquisition system

Figure 4. Mental effort acquisition system

Eye movements will be recorded in video using adedicated optical tracking located near subject's headthat detects the subject's retinal position (Figure 5).This data is related to the viewing direction and it isvery important because it permits to determineattention and distraction rates.

Optical sensor

Video recorder Monitor

Data acquisition system

Figure 5. Eye movements recorder equipment

Both RSME and NAC equipment are synchronizedwith the simulation experiment by a developedinterface module (Figure 6).

RS-232

Figure 6. Synchronizing the acquisition equipment with simulation experiment

This module receives information about the simulationstage from the graphical workstation through astandard serial connection (RS-232) and outputs therequired synchronization signals. The synchronizationallows specialists to relate the outputted data with thesimulated traffic events.

Learning measures will be the performance measures:the number of errors committed• reaction times• time-to-line crossing• control of the driving path• driving speed• number of collisions

These measures are obtained by processing theinformation that the simulator reports in real-time.

This study reflects the relation between the drivingsimulator use and learning process in these courses.The performance results from these initial courses arepresumed to be different from each other, even with thespecified differences in course duration. An adaptationof each course curriculums must be done usinginformation from the results of this preliminaryexperience. This adaptation is important becauseperformance results from different courses can only becompared if they have some similarities.

The final experiment of this study will use the samemethods but with differences on the number of usedsubjects, the number of groups and courses, and thecurriculums of the driving courses.

A second study will be done on the validation ofdriving learning assessment. Driving teachers andexperts will be asked to analyze video records from thefirst study (images recorded from the driver point ofview), and to score the drivers performance with anevaluation sheet used at driving schools. The objectivedata from learning measures gathered at the first studywill be then matched with the subjective evaluationfrom experts. This second study will allow us to

analyze the value of subjective reports from drivingteachers on topics such as vehicle control, skills oflearning drivers in unexpected situations, learningimprovement along driving sessions and agreement ofexperts in the assessment of driving skills.

As results of these studies, we expect to determine theinfluence of the simulation realism over the learningprocess, and to understand how and when the drivingsimulator should be used to maximize the drivingeducation’s productivity.

6. Conclusions

This paper shortly presented the architecture of DriS, arealistic driving simulator. DriS integrates a typicalcockpit of a real car, a SGI Onyx Reality Engine 2graphical workstation and a large projection screen. Italso uses a PC for sound generation, and another PC tocontrol and monitor cockpit sensors and signals. DriSis being developed to study driver behavior, roadperformance and driving education methods.

Experiments being prepared to the evaluation ofdriving education methods were also presented. Theseexperiments will use data acquired from a large set ofpersons without driving license. Each person will besubmitted to a driving course that may include drivinglessons in a real car and / or in the driving simulator.The realism used in the simulation will be adjustedbetween the classical non-interactive, film basedtraining and the full realism achieved by our drivingsimulator. Each driving education method will beevaluated by comparing driver’s competence beforeand after the applied course.

We expect these experiences will allow concludingabout the required level of realism that a drivingsimulator must achieve in order to be used in drivingeducation processes and to understand how and whenthe driving simulator should be used for betterperformance in the education process.

7. References

[1] Salvador Bayarri; Fernandez M.; Martinez M.,Virtual Reality For Driving Simulation,Communications of the ACM, May 1996

[2] J. MIGUEL LEITÃO; A. COELHO; F. N.FERREIRA; DriS – A Virtual Driving Simulator;Proceedings of the Second InternationalSeminar on Human Factors in Road Traffic,ISBN 972-8098-25-1, Braga, Portugal, 1997

[3] Salvador Bayarri; M. Fernández, Design &Validation of Traffic Models in DrivingSimulation, Symposium on the Design andValidation of Driving Simulators, Valencia,Spain, 1996

[4] Olivier Alloyer, Esmail Bonakdarian; JamesCremer; Joseph Kearney; Peter Willemsen,Embedding Scenarios in Ambient Traffic,Proceedings of DCS'97 (Driving SimulationConference), Lyon, France September 1997

[5] J. Miguel Leitão; A. Augusto Sousa; CarlosRodrigues; Jorge A. Santos; F. Nunes Ferreira;A. H. Pires da Costa, Realização deExperiências num Simulador de Condução,Psicologia del Tráfico y la Seguridad Vial / IICongresso Iberoamericano de Psicología,Madrid, July 1998

[6] António C. Coelho; J. Miguel Leitão; F. NunesFerreira, Síntese de Som 3D em Ambientes deRealidade Virtual – Aplicação a Simuladores deCondução, Proceedings of the 8º EncontroPortuguês de Computação Gráfica, Coimbra,Portugal, 1998

[7] Paulo Noriega; Jorge Santos; Carlos Rodrigues;Pedro Albuquerque, Vehicle’s MotionDetection: Interaction with three Kinds of RoadPavement, Proceedings of the SecondInternational Seminar on Human Factors inRoad Traffic, ISBN 972-8098-25-1, Braga,Portugal, 1997

[8] J. Miguel Leitão; F. Nunes Ferreira, AgentesAutónomos em Ambientes Artificiais,Proceedings of the 8º Encontro Português deComputação Gráfica, Coimbra, Portugal, 1998

[9] A. F. Sanders, Simulation as a Tool in theMeasurement of Human Performance,Proceedings of the Ergonomics Society AnnualConference, Leeds April 1990

[10] James Cremer, Joseph Kearney, Yannis Papelis,Driving Simulation: Challenges for VR

Technology, IEEE Computer Graphics andApplications, September 1996, pp. 16-20

[11] Lawrence H. Frank, John G. Casali, Walter W.Wierwille, Effects of Visual Display and MotionSystem Delays on Operator Performance andUneasiness in Driving Simulator, HumanFactors, 1988, 30(2), 201-217

[12] Pieter Padmos, Maarten V. Milders, QualityCriteria for Simulator Images: A LiteratureReview, Human Factors, 1992, 34(6), 727-748

[13] Steven B. Rogers, Walter W. Wierwille, TheOccurrence of Accelerator and Brake PedalActuation Errors during Simulated Driving,Human Factors, 1988, 30(1), 71-81

[14] Michael L. McGinnis, George F. Stone III,Measuring the Effectiveness of Simulation-Based Training, Proceedings of the 1996 WinterSimulation Conference, pp 934-938

[15] J. Miguel Leitão; A. Augusto Sousa; F. NunesFerreira, An Image Generation Sub-system for aRealistic Driving Simulator, Proceedings of the1999 International Conference on ImagingScience, Systems, and Technology, Las Vegas,June 1999

Scientific Visualizatio n as a Tool for Power Engineering Education

Pavel Slavik, Department of Computer Science and EngineeringFrantisek Hrdlicka, Department of Fluid Dynamics and Power Engineering

Czech Technical UniversityKarlovo nam. 13, 121 35 Praha 2

Czech Republice-mail: slavik@cs.felk.cvut.cz, hrdlicka@fsid.cvut.cz

Abstract

This paper deals with the application of scientificvisualization methods in the field of power engineeringeducation. Behavior of various parts of a power plantis simulated and the results of these simulations arevisualized. The visualization of processes that run in apower plant gives the students an opportunity tounderstand better the nature of these processes. Thisapproach solves one of the main problems in powerengineering education: the students usually have nochance to see various types of power plants in practice.The existence of models and their correspondingvisualization allows students to perform experimentswith various devices and in such a way acquire a muchdeeper knowledge about the subjects taught.

Keywords: power engineering, power plant,visualization, particle models.

1. Introduction

Production of electrical energy plays an important rolein national economies around the world. Therefore it isnecessary to pay attention to the education of highlyqualified professionals in the field. They should be ableboth to design and to run power plants of variouskinds. Education of these experts is very costly and,moreover, very complicated. One of the crucialproblems is the lack of practical experience thestudents can gain during their study. Only practicalexperience can help the students deeply understand thenature of physical processes that run in a power plant.The traditional way of education is mostly based onmastering the theory of these processes, which createsa good base for their later deep understanding inpractice. In some cases real models of some parts ofpower plants are used, where the students can performexperiments. This approach has several disadvantages -first of all these models are rather costly. Alsoparameter settings during experiments are usuallyaccompanied by various problems.

The solution to the problems stated above is to usecomputer-based models of power plants. Currentlythere exist commercially available models that offerthe possibility to perform virtual walkthroughs in apower plant. Such a walkthrough can give studentssome idea about the configuration of a power plant andalso about the size of single appliances in a powerplant. Performing such a walkthrough, students can geta feeling about mutual relations between single parts ofpower plant, which allows them to understand globallythe process of electricity production. Nevertheless, thisapproach does not say anything about the processesrunning inside single parts of a power plant. Thesolution to this problem is to create models ofprocesses running in power plants. These modelsshould have both the simulation part and thevisualization part.

2. Models of power plant components used inpower engineering education

At Czech Technical University in Prague a project hasbeen running for several years where these models aredeveloped and implemented. Their applicability toeducation has been extensively tested. The aim of theproject is to develop a set of models (andcorresponding program modules) that would allow usto simulate various modes of operation of each part ofa power plant. These modules should be of a specialnature as it would be important to combine thesemodules into a functional model. For example: it ispossible to assemble complex models of various powerplants types that are characterized by different types ofboilers, different types of coal transport systems, etc.Such an approach can significantly improve thequality of education, as a variety of single power planttypes is usually not available in one country. Only bymeans of these models is it possible to acquaintstudents with all possible aspects of design andoperation of power plants in a very general way.The project as a whole is a large scale project whichmeans that the final state will be reached within severalyears. Nevertheless, several modules currently exist

which are used independently in education with a highdegree of success. This paper will deal with the currentstate of the art of the project mentioned. The simulationand visualization modules use traditional visualizationtechniques (mostly based on the use of particlesystems). Some simulations and visualizations areunique because of the very specific nature of theprocesses investigated. It was necessary to developspecial simulation and visualization techniques thatallowed us to investigate the behavior of theseprocesses - professional simulation and visualizationsoftware mainly did not cover specific requirementslinked up with the processes simulated.As the system is targeted to education, there are somespecial features in which this system differs fromtraditional simulation and visualization systems used inthe given field. The traditional techniques mainly stressthe accuracy of modeling and visualization. Thisapproach results in a very time consuming calculationsthat give results after very long time. Our approach hasbeen slightly different in some cases: to use simulationand visualization techniques that have lower accuracyand thus lower requirements for computational time.This offers very flexible feedback to students as theycan see the results immediately (mostly in real time). Insuch a way students can get a feeling about thedynamics of the process simulated (and visualized).The next step in using the system developed could bethe use of professional specialized software (likeFLUENT) and comparing the results obtained from oursystem with the results obtained by a professionalsystem. A good exercise would be to modify existingalgorithms used in our system in order to improve thematch with the results obtained by professionalsoftware. Such an approach is not always possiblebecause some models in our system are quite uniqueand they have no counterpart in professional systems.In this situation students compare the simulated resultswith real situations. They can compare the model andthe reality and again explain why these differencesexist. The experience gained could be used in thefuture for system improvement.

3. Description of existing modules

The already existing modules cover the most importantparts of the electricity production pipeline. This meansthat the student using all these modules can get acomplex idea about mutual relations between singlestages of electricity production.

The modules available are as follows:

- Heat transport in nuclear reactor cooling systemIn this case the data obtained by simulation of nuclearfission (here a commercial software product has beenused) are visualized by a special module. This moduleis a good example of a situation where sophisticatedcommercial visualization software exists. The problemis that this commercial software runs on powerfulworkstations which means that this software is notsuitable for education on a larger scale. Our softwareallows us to visualize heat flow and temperaturedistribution. The solution was based on the use of finiteelements for the reactor body description andvisualization of temperature and heat flow within theseelements.

- Combustion processes in fluidized bed boilersThe problems related to fluidized bed boilers are rathercomplex. In the module a particle system was used.The particles represented coal particles during thecombustion process. The attributes of particles changedaccording to the degree of their burning out. Thevisualization is then quite simple. Due to the simplicityof the particle model it is possible to generate animatedsequences in real time and in such a way to observeand investigate the dynamics of the combustionprocess (Fig.1). Comparison has been made withsimulations and visualizations made by means of amathematical model that was based on the theory offluid dynamics.

Fig. 1 Visualization of various combustion modes

This model provides a more general approach to thestudied phenomena. Nevertheless it is computationallyexpensive and each change of parameters requires it toperform new simulations (that are computationallycostly). The particle model implemented allows us toperform changes of parameters dynamically withimmediate visualization. Students can followimmediately the influence of the parameter changesjust being made. This feature means the particle modelis superior in terms of its use in education.

- Combustion processes in vortex furnaceThe experience gained with the preceding boilersystem has been used to create the model of acombustion process in a vortex furnace. One of theprincipal parameters of this boiler system is theconfiguration of an air velocity field that influences thetrajectory of coal particles in the furnace. As in theprevious example it is possible to set parameters of theboiler interactively by means of a properly designeduser interface.Also in this case the use of the particle model bringsthe same advantages as in the previous case. Due tosome optimization in the program it is possible to workwith several thousand particles (Fig. 2) at the sametime (and visualize them). Such a large number ofparticles gives very extensive information about thesituation in the furnace during the course of simulation.

Fig. 2 Visualization of combustionprocess in vortex furnace

Fig. 3 Visualization of the number of collisions

Fig. 4 Visualization of the coal transport in the given pipe configuration (including collisions)

Fig. 5 Distribution of pollution from three sources

- Coal transport systemThe system allows us to investigate behavior of coalparticles during their transport in a coal dryer. Theconfiguration of pipes the dryer consists of (Fig. 4),determines the extent of collisions of coal particleswith the inner surface of the pipes. The number ofthese collisions determines the level of damage to thetransport system. Another factor that influences thecoal transport are parameters of coal like its humidity,granularity etc. The system allows the user to setinteractively all these parameters and in such a way tosimulate the coal transport process. The result ofsimulation and visualization is a picture where thedistribution of collisions is shown.The Figure 3. shows the visualization of the number ofcollisions of coal particles with the inner surface of thepipe through which the coal is being transported.

- Moving granular bed gas cleanup filtersThe mentioned cleanup filter has been modeled. Themodel describes the behavior of granules (they havespherical form) - how they move in the filter during thetime period of filter operation. Also here the processhas several parameters that influence the efficiency ofthe pollutant separation. The user can experiment withvarious parameter settings and in such a way betterunderstand the nature of the gas cleaning process. Thisproblem as a whole is currently under intensiveinvestigation and includes calculation of powers actingon single sphere in various positions inside the filter.Another feature under investigation is the calculationof friction processes between spherical granules insidethe filter. The simulation of the granular bed gas filteris the most complex one of all problems investigatedand mentioned in this paper.

- Simulation and visualization of pollutantsdistributed from a stack in a power plantA model has been implemented that describes thepollutant distribution in an area near the power plant.The terrain profile is taken into account, and also themeteorological situation and characteristics of thepower plant (and the stack in particular) are taken intoaccount (Fig. 5). It is possible to simulate the influenceof several power plants in a specific area where theshare of pollution caused by each single power plant iscalculated. The system has also been used in the coursewhere the suitability of a location for a new powerplant was tested.

4. Implementation

Most of the modules described were implemented inthe MS Windows environment. The fact that most ofthe modules are available on the MS Windowsplatform is very important as these modules can bewidely used due to wide availability of this platform.There are no specific demands on hardware.Nevertheless in the case where some dynamical effectsare simulated, the appropriate computational power isof great importance. Great attention was paid to thedevelopment of a proper user interface keeping in mindthat the modules could be used also by users who donot work in the field of computer science - so theirknowledge about using computers could be defined asa casual user level.

5. Evaluation

From the examples given above it is possible to seethat the modules already in use cover in cover severalsingle stages of the electricity production. In such away it is possible to see how the situation in one stageinfluences the situation in the next one - and also itsglobal influence on the electricity production process,including the consequences for the environment.By means of these program modules extensiveteaching materials were prepared. Thanks to theflexibility of parameter settings in each of thesemodules, it was easy to generate a wealth of teachingmaterial that was used both in lectures and in seminars.As an example it is possible to give the course”Combustion and Boilers” where the visualization ofthe combustion process allows students to understandthe mutual influence of all parameters that take part incontrolling the combustion process. In such a way it ispossible to determine ill-suited combinations ofparameters (these combinations should be avoided inpractice). The dynamics of the combustion is shown inthe system by means of suitable visualizationtechniques.In the framework of another course called”Environmental Problems of Power Generation” themodel of plum rise dispersion was used. Also in thiscase student could get a much better sense of the natureof the problem. Students could better feel the influenceof each single source of pollution to the environmentalsituation in the given area. This means, for example,that the influence of the planned power plant can bedetermined in advance by means of the modelimplemented.Some other models mentioned have been used to alimited extent (up to now) in the framework ofseminars (e.g. for PhD students) and in specializedlabs. This approach allows us to prepare the use ofexisting modules for regular courses. A very important

benefit is that the simulated results can be in manycases compared with the real data obtained in somecooperating power plant. This creates a very intensivefeedback between the research at the university andpractice. The environment created for universitystudents in this way is, according to our experience,very stimulating and students get more involved bothin research and in study.Because the usage of these materials is quite new, thereis still not enough data for the qualitative evaluation ofthe impact of this new approach on the quality ofeducation. An evaluation strategy will be developed inthe framework of future work. Nevertheless it ispossible to say that this new approach led to a muchbetter understanding of processes running in a powerplant, which resulted in better results on exams insingle courses.

6. Conclusion and future work

Due to the pilot nature of the program modulesimplemented, there is still a lot of work to be done inorder to improve them. Moreover there are many partsof power plants of various types that should also becovered by corresponding modules. As an example, wecan give modules for the heat transport andtemperature distribution in boiler furnaces of varioustypes, ash particle transport through boiler convection

heat exchanges, etc. Some future developments shouldbe the result of consultations with experts working inpower plants in order to establish the real needs forsome new models.

References

[1] Slavik, P., Hrdlicka, F., Pavlis, T., Vlcek, P.:Visualization of the Dispersion of AirborneEffluents,In Proceedings of 7th EurographicsWorkshop on Visualization in ScientificComputing, CTU Prague, 1996, pp.153-169

[2] Hrdlicka, F., Slavik, P.: Dynamic Models ofthe Lignite Transport in Dryer of PowerPlant Boiler, To appear in Proceedings of24th International Technical Conference onCoal Utilization & Fuel Systems,Clearwater, Florida, 1999

[3] Kubelka, F.: Simulation of Moving GranularBed Gas Cleanup Filter, Student project,CTU Prague 1998

[4] Faltyn, R.: Simulation and Visualization ofCombustion Processes in Vortex Furnace,Master Thesis, CTU Prague 1999

[5] Rais, D.: Visualization of CombustionProcesses in Fluidized Bed Boilers, MasterThesis, CTU Prague 1997

Training in Computer Graphics for Entertainment Production:What Future TDs Need to Know

Jacquelyn Ford MorieRhythm and Hues

jacki@rhythm.com

Abstract

Training grounded in the theory and soundconcepts of any discipline is widely recognized asthe most appropriate educational practice whichserves students well beyond their graduation. Butin an increasingly competitive world students canbe given a jump-start into a career by focusingtheir education towards an understanding of theknowledge set they will need for certain types ofcareers that use that discipline. Computergraphics is an especially rich area withinComputer Science or Engineering departments, inthat a large number of careers make use of thesame basic concepts. However, careers incomputer graphic areas almost always requireother skills sets of its potential employees, as mostare interdisciplinary professions where computergraphics forms but one aspect of the actual job.Examples of this are numerous and range fromarchitecture to CAD/CAM to medical applications.The specific career that I will discuss as anexample of training for real world applicationshas been a highly sought after one for manycomputer graphic students in the last decade —that of technical director for entertainmentapplications such as film, television and games. Inspite of adequate computer science theory andconceptual training, many graduates are notaccepted by this industry because of a lack ofunderstanding of what is expected, and acorresponding lack of additional relevant skills.This paper explores some of these desired skillsand serves as a basis for discussion for howeducating students towards an understanding ofskills necessary for many computer graphiccareers can be included in a traditional computerscience or engineering program.

Introduction

One of the faster growing segments of the jobmarket for graduating Computer Science (CS)majors over the past six years has been that of thetechnical director (TD) or software specialist forthe entertainment industry. Students to fill these

needs within the industry have been in shortsupply throughout this time. Many companieshave had to look overseas or to related industriessuch as aerospace or other engineering disciplinesto find the talent they required during what hasbeen an unprecedented growth period withindigital entertainment. This hiring burst of recentCS graduates, at all levels, has brought with it arange of higher salaries, which, in addition to thehigh-profile nature of the jobs (from screen creditsto the glamour of Hollywood), has left a shortageof these same students to populate positions inmore traditional fields, such as defense, orCAD/CAM applications.

Traditional Computer Science education, while itprovides students with the programming skillsthey need and the ability to approach advancedproblem solving in a logical fashion, does little toprepare them totally for new and distinctivecareers such as those in digital entertainment.Even with the best Computer Science education aspreparation, studios and production companieshave found that there is an additional level ofeducation that these incoming hires often need, nomatter what their specific job function within thecompany.

The types of knowledge lacking in these hires isusually not part of, nor even considered within, atraditional CS program. It consists, in part, ofcompetence in areas of communication andmanagement, artistic expertise, and knowledge ofproduction itself and the way computer tools areused within the various types of productionpipelines. Can these areas be folded within atraditional CS curriculum? Should they? Thatdepends. If the task of training students is toprepare them for future jobs, then perhapsconsideration should be given to methods withwhich to prepare them better. Several suggestionson how to accomplish this for the entertainmentindustry are explored, hopefully sparking a dialogleading to an enhanced computer scienceeducation.

The Rapid Rise of Digital EntertainmentJobs

In 1992 James Cameron had a huge box-officesuccess with a sci-fantasy film called Terminator2: Judgement Day. This film was significant inthat it could not have been realized withoutextensive use of computer generated imagery(CGI) to create what was essentially a maincharacter in the film. More importantly, this filmwas a watershed in a long range economic way —it gave studio executives a level of confidence thatthis technology could actually be used in filmproduction, without exceeding the economicconfines of time or budget. The year beforeTerminator 2's release (1991) there wereapproximately 2000 people making their living asdigital animators of some sort in the entireindustry. A year after T2's success (1993), therewere close to 4500! And today there are well over10,000 digital animators and artists, includingtechnical directors employed in the business.[1]

Digital entertainment has become a significant—not to mention exciting—job market for computerscience graduates searching for their first job.

This rise in numbers of jobs has been aninteresting phenomenon, partly because the jobmarket increased so fast that few schools wereready to meet the challenge of training students foremployment in these new jobs. This led to a needfor production companies to institute their owninternal training programs to help new hires bridgethe gap between the skills they had and those theyneeded to be successful. These programs rangebetween a few days (for someone with previousproduction experience who only needs to get up tospeed on minimal proprietary software) to a fullyear for recent graduates who have much to learn.The cost of these internal training programs hasnot been insignificant, but companies simply hadno choice. If the dollars spent over the last fewyears in this area were to be added up theeconomic impact would be seen to be profound.And since the profit margins, especially for smalleffects companies who do not own their owncontent, are often in the single digits, internaltraining programs are not likely to survive ifcompanies can find any way around them. This isespecially likely given the current state of theindustry, which includes a number of significantlayoffs and company closings, thus puttingexperienced people out into the job market againin large quantities.

While it is true that the previous demand for moreand more hires is currently reversing, it must berealized that the entertainment industrytraditionally experiences ups and downs. Thiscyclic nature of entertainment is another importantreason to provide students with as muchknowledge as possible while they are in school.Not only will those with more knowledge be thelast to be let go in the slow times, they will bepreferred hires when the pace picks up over otherentry-level hopefuls who would need significanton-the-job training to come up to speed.

What They Need To Know

The jobs of technical director and entertainmentsoftware developer are not widely understood.Even within the entertainment industry the samejob title can encompass a wealth of differentexpectations, with an increasing trend towardsspecialization. For example, a recent recruitmentad for Industrial Light and Magic, a division ofLucasfilm, Ltd., listed the following specializedjob titles, all of which might be lumped togetherunder the "TD" category at a smaller company:Digital Matchmovers, Envelopers, CGCommercials Artist, CG Software Engineers, andSenior Level Technical Directors. The general ruleis: the smaller the "house" the more hats anemployee will wear.

The skill sets necessary to be a successfultechnical director or software developer in theentertainment industry are also fairly wide range.While any given job may not absolutely require allthese skills, knowledge or understanding of themwill significantly improve the efficiency withwhich the job is done, thus maximizing the valueof the employee to the company. The followinglisting of additional skill sets beyond those oftraditional computer science programs is drawnfrom the most common requests and observationsreceived from supervisors during production.

Communication, Social and Management SkillsThis is an extremely important set of competenciesand often the ones that have the most bearing onan employee's eventual success or failure. Theseskills include: how to work with artists, andproducers, and directors; understanding thedemands and rigors of production schedules andhow to work within them; coping with extremelyshort research and development times; keepingconfidentiality; and accepting criticism in aconstructive way, by using it as a tool to improve

the quality of the work It is often difficult for newemployees to work long hours, not talk about whatthey are doing with their friends, and realize thatthe director is always right, even as they redo ashot for the twentieth time at one o'clock in themorning. And technical directors often requirestrong management skills to lead those workingunder their direction, including knowing how toinspire a team of people facing the above pressuresto push their artistic and creative limits tomaximum effect.

Artistic SkillsAesthetics! The best TDs are those whounderstand artistic fundamentals and havedeveloped a strong sense of aesthetics. At aminimum, TDs need to know how to value the artskills of those they work with, and collaboratewith artists for the best results. But understandingthe art themselves is best. A fundamentalcomprehension of design concepts provides asignificant advantage for TDs. Most all TDs wouldalso benefit from knowing color theory, and morecolor theory. It is difficult to find experts in thiswho understand the whole color picture, as it istaught so differently within different disciplines,when it is taught at all. CS students need tounderstand about color psychology, color spacesthat artists deal in, and color phenomena as well asgamma and wavelengths. Lighting is another areain which TDs need artistic training. Lighting TDsare in extremely short supply, typically becausethose who understand how to use the lightingtools, such as Renderman™, do not usually havethe co-requisite skills for manipulating light inaesthetic ways.

Learning and Research Skills

In addition, TDs will consistently be called uponto invent something that has never been donebefore. The best TDs are those who can makeingenious and intuitive leaps that bring unseenvisions to life. They need to understand what hasgone before and how to put seemingly disparatepieces together. And they will always be expectedto keep up their skills, learning or writing newsoftware on very short time frames.

Specific Job SkillsSome specific topics that TDs should knowinclude: scripting and expression writing to createor control images; modeling complex, organicobjects and how continuity is kept across such

surfaces; enveloping, or developing the final"skin" around an animated character; riggingcharacters for animation, which is setting upintuitive controls for animators to use; proceduralmodeling using such tools as particles or implicitsurface techniques; detailed shader and texturewriting, which creates the final look of an objector character; pixel mathematics, which are usedextensively in compositing operations; andworking with 2D and 3D spaces and theirinteractions, especially in techniques such as 3Dtracking. It is also imperative for TDs tounderstand how film works—how it is exposedand scanned into the computer as well as how tocorrectly output digital images to it. And last butnot least, an understanding of the entire productionprocess is extremely helpful. Knowing how andwhy certain decisions are made for a particularscene before it ever gets to the point where thedigital manipulations occur can simplify the TD'stask. Add to these areas an understanding of howto decide which tool is best for which job, sinceCG may not always be the best choice, or when tochose one CG tool over another. For example,modeling an asteroid is easier using simplegeometry and a clever shader rather than a modelwith ten thousand polygons, and blowing up abuilding might be simpler to achieve with apractical, miniature model than with CGI.

Suggestions on How To Do This

Tying the appropriate computer scienceknowledge to practical applications is much easiertoday than it was twenty years ago. Howevercomputer science education has not changed toreflect this as much as it might have. CS has a setof core classes that are certified and endorsed byvarious academic and computer organizations.Contrast this with the multitudes of "ComputerGraphics" programs within art, design,communication, architecture, film, tv, or mediadepartments that have been created over the pasttwenty years. These more artistically-inclinedprograms have very little in the way of standard orendorsed "core" classes. These programs havebenefited, however, from a great deal of advicefrom studios and entertainment companies, andhave modified their programs accordingly, withmore or less success. They still, for the most part,lack training in the more technical areas ofexpertise needed for the optimum success in theentertainment industry. Both types of programsproduce students who do not have all they need.

It is tough to create a truly interdisciplinaryprogram within most university or collegeenvironments. Not only are there scarce resourceswhich often leads to departments protecting theirdomains, there really isn't enough time in a typicalfour year program to implement the total numberof classes that would create the ideal graduate forthis industry. Nor is there anything remotely likean apprenticeship program for the digitalentertainment field (though this is currently beingpromoted to the State of California). So how canwe provide ways for CS students to become moreprepared for jobs in this arena, without diluting thecore education they need?

One obvious way is to have more input from theentertainment industry. This increased interactionmight lead to additional resources such asentertainment-sponsored training materials orinnovative elective classes taught on video, theinternet or via video conferencing. On-line guestlectures from top TDs at a variety of productioncompanies could be collected and distributed to anumber of participating schools. (SIGGRAPHcould sponsor this as a Special Project.) Onlinementoring or critiquing is another option, where anindustry professional nurtures, reviews or critiquesstudent projects using student web pages.

New teaching methods within CS is anotherpossibility, such as special topic classes that areteam-taught and co-listed between departments.Dr. Dave Ebert and Dan Bailey are teaching justsuch a class at UMBC in Spring 1999, called "Artand Science in Computer Animation", with ten artstudents and ten CS students working oncollaborative computer animation projects. Therecould also be more emphasis on availablematerials or publications, such as Cinefexmagazine, or various special television broadcasts,which would be valuable for those studentsinterested in these careers. This can be donethough seminar classes or through career resourcedepartments. There are many points ofintersection, especially for advanced or seniorlevel classes that allow students to bring theircomplementary skills together in production-likeprojects, thus learning from each other. Even a fewsuch classes or techniques as described abovecould go a long way towards providing studentsbetter preparation for careers in the digitalentertainment industry.

[1] "The Visual Effects Industry Needs a NewProduction Model: Why Titanic Is the 2001 of OurTime", by Ray Feeney, Chuck Spaulding, andKevin Mullican, Millimeter, November 1988, pp.65-71.

Human-Computer Interaction and Cognitive Psychology in VisualizationEducation.

Thomas T. HewettDepartment of Psychology/Sociology/Anthropology

Drexel UniversityPhiladelphia, PA, USA

hewett@drexel..edu

Abstract

A curriculum designed to educate those who design,develop, and deploy scientific visualizations and/orvisualization tools needs to contain a strongcomponent of both human-computer interaction andhuman cognition. The argument centers around twoobservations. The first is the rapidly increasingbreadth of application for computer-basedvisualizations and the tools for creating them. Thesecond observation is that increasingly problemsaddressed using visualizations involve representingabstract objects, entities or relationships for whichthere is no obvious direct physical analog. Both trendspoint towards an increased need for understanding andincorporating visualization needs of an increasinglydiverse set of users. In addition the trends towardsincreased diversity of users and the abstractness ofrepresentations being created for their use virtuallydemand an increased understanding of psychologicalnature of visualizations and their effects.Keywords:human-computer interaction, cognitivepsychology, visualization education

1. Introduction

Instances of scientific visualization currently rangeover many domains of human knowledge. Forexample, computer-based visualizations have beenused to expand and contract time and space in waysthat range from the modeling and rotation of entiregalaxies to the enlargement and manipulation ofmodels of molecular structure. Similarly,visualizations have made the invisible visible in waysthat allow such things as viewing the dynamicmovement of the valves the human heart and magneticresonance imaging support for microsurgery. Inaddition, visualizations have made the difficult easy asin the case of software that graphs equations forcalculus students or that enables a researchmathematician to generate and manipulate complexsurfaces as an aid to solving equations.

Currently, the ease with which computers track, record,and display information is leading to a flood of new

uses of information visualization. Consequently, it iseasy to envision even more diverse applications ofcomputer-based visualizations in the near future as ourknowledge of how to control the computer andvisualization tools grows. For example, given thatwork is already taking place in these areas describedbelow it is relatively safe to predict that in the next fewyears we will see an increasing use of and demand forinformation visualizations and tools for use in suchapplication domain areas as:

1) Biomedical engineering of human healthand performance through the creation of new devicesfor sensing and imaging of information which can beused in the design and development of new prostheses(e.g., stents to reinforce weak blood vessels, artificiallimbs with improved functionalities, etc.) and newvisualizations of complex patient data (e.g.,combinations of vital signs and indicators beingmonitored on a patient in the hospital Intensive CareUnit.

2) Environmental science work engaging suchdiverse problems as automobile traffic control toreduce average trip time and auto-emissions;atmospheric or oceanic modeling of diffusion andchemical transformation of various pollutants;modeling the urban infrastructure to support planingand resource allocation.

3) The utilization of geoinformation systemsin coping with such problems as geolocation and way-finding, public and private land use decisions, wildlifemanagement, and police detection and control ofpatterns of criminal activity.

In addition, it is also easy to imagine that computer-based visualizations will play an increasingly largerrole in future human use of information technology.The complex set of technological, environmental, andsocial problems facing the human race virtuallydemand that information technology (includingcomputer-based visualizations) be used to support life-long career activities in knowledge workers from manydomains. If correct, this prediction suggests that wealso need to be concerned with the development of awide range of strategies for reskilling and regenerationof professional careers. Many of these strategies for

continuing professional education will, of necessity,come to depend more and more on the development ofcomputer based visualizations which offer us new andfruitful representations with which to think aboutproblems.

2. Why Human-Computer Interaction andCognitive Psychology?

In all of the cases described above, visualization of theabstract will come to play an increasingly greater role.In other words, the problems to be addressed usingvisualizations will increasingly involve representingabstract objects, entities or relationships for whichthere is not an obvious direct physical analog [1].Considering the wide range of technical subjects andknowledge areas which will contribute to the solutionsof the types of problems described above, how are weto sort out which blocks of knowledge should be partof the education of developers of visualization toolsand visualizations? Clearly such people will need towork in multi-disciplinary teams in order to call uponthe contributions which can be made by specialistsfrom other disciplines. Given all the technical materialthat a student of visualization will need to learn, andgiven the wide range of specialists with whom thestudent will eventually interact, why would anyonewant to argue that human-computer interaction (HCI)and cognitive psychology are critical knowledge areasto be included in a curriculum for visualizationeducation?

2.1. One Answer to the Question.One simple answer to the question of "Why HCI andcognitive psychology?" is that there is alreadyagreement about the importance of both topics tocomputer science education in general. For example,the report of the ACM/IEEE-CS Joint Curriculum TaskForce [2] clearly identifies human-computercommunication as one of the nine core areas ofcomputer science. In this core topic the report authorsinclude both computer graphics and human-computerinteraction as related topics. Furthermore, the authorsof the report point out that cognitive psychology is animportant supporting discipline for three of the ninecore areas of computer science, including artificialintelligence and human-computer communication.

2.2. More about HCI.Beyond the fundamental natural affinity that existsbetween computer graphics and HCI, a moresubstantive argument for the importance of an HCIcomponent to visualization education can be found intwo simple observations. First, most computer-basedvisualizations or visualization tools are created with theintention of supporting some sort of human interactionwith the visualization and the computer which creates

it. Second, there is already in existence a reasonablyclear delineation of the conceptual structure andknowledge modules which are central to the topic ofHCI [3]. These resources have already proven usefulin guiding the organization of major textbooks in HCI[e.g., 4] and continue to be of interest (e.g., during thelast 12 months the report web site recorded over 9,000hits, even though the report was published in 1992).Consequently, this existing catalog of knowledgemodules makes it possible to explore and define thecomponent aspects of HCI which should be consideredpart of an education in visualization technology.

2.3. More about Cognitive Psychology.Substantive arguments for inclusion of the study ofcognitive psychology in a curriculum on visualizationderive from the fact that computer-based visualizationsand tools are typically designed to make it possible forsome sort of knowledge worker to achieve a set ofwork related goals. Typically, these goals involvedeveloping the solution to a complex problem, or theyinvolve the creation of new knowledge. Thus,designers and developers of computer-basedvisualizations and visualization tools need to be able toground their designs in a knowledge of the end-usertask or situation for which the visualization is beingdesigned. Increasingly these visualizations and toolswill require dealing with abstract relationships that areuseful in enabling user’s to think about their work.

As illustrated in [5], external cognitive representationscan be powerful tools with which to think. Focusingon information visualization, Card, Mackinlay andShneiderman [6] suggest six major ways in whichvisualizations can be used to amplify human cognition.Visualizations can increase the external memory andprocessing resources available to a user. They canreduce the user’s need to search for information. Theycan enhance the user’s ability to detect patterns in dataor events. They can facilitate the drawing of someinferences through direct perception of the informationrather than through more complex cognitiveprocessing. They can facilitate the monitoring ofchange in large numbers of potential events. Finally,visualizations can be used to encode information in amanipulable medium.

The need to understand abstract relationships which theend-user is attempting to capture necessarily entailsthat the designer of a visualization, or of the toolsneeded to work with information visually, should beable to collaborate with the end-user in identifying thedemands of the situation and in developing aknowledge representation appropriate to the way theend-user thinks, or needs to think, about theinformation being visualized. In other words, thedesigner has to develop an understanding of theproblem being solved, why it is a problem in the first

place, and what the constraints that determine anappropriate visualization.

However, for individuals educated only in computerscience the problem of understanding the user's taskoriented needs and goals can constitute a major barrierto development of appropriate tools, etc. For example,as pointed out in [7] and confirmed by the work ofHarrison, Hewett and Perline [8], visualization ofmathematics has some special features that requirenovel visualization techniques. Developers of manyexisting visualization packages have tended to viewtheir task in a holistic manner which does not takethese special needs into account. What has matteredfrom the perspective of the developers of existingvisualization tools is the accuracy and speed of displayof the whole model or scene. Consequently renderingis done in a buffer storage and is invisible to the useruntil completed. However, it turns out that in this caseaccuracy and speed aren’t just irrelevant to manymathematicians, accuracy and speed may becompletely contrary to the needs of mathematicians.For example, a visualization of the entire surfacegenerated by an equation is not necessarily what isrelevant to the mathematician exploring it. Whatmatters even more is the evolution of the curve throughanother variable (e.g., time) that creates the surface.This necessitates a set of visualization tools and avisualization that can show the surface, the curves andthe growth of those curves into the entire surface.

Another difficulty with traditional computer scienceeducation lies in the fact that our understanding ofhuman behavior is, on the whole, limited to heuristicknowledge, whereas, computer science educationtypically deals with algorithms. Computer scientistsare educated to deal with issues such as computabilityand completeness. Cognitive psychologists, because ofthe nature of their subject matter, are forced to spendmuch of their time working with heuristics, workingwith the conditionality of certain kinds of knowledge,and investigating knowledge representation. Indeed, alarge measure of what has been learned in cognitivescience in general is conditional knowledge whichdepends upon the circumstances in which informationhas been collected and which can only be characterizedin the form of heuristics and conditional statements.

2.4. An Example of a Failure to UnderstandCognition and Conditionality.Although not based in an instance of scientificvisualization, this example illustrates how afundamental misconception about cognition mightnegatively impact the design of a visual representationof information. It also illustrates an importantrelationship between what has been referred to asknowledge in the head and knowledge in the world [5].

At a University which will forever remain namelessthere are classes in which students are taught aboutHCI. In some classes, students are given the followingdesign guideline: “The Macintosh interface is a badinterface because there are menus with more thanseven items,” the implication being, “Do not designinterfaces such as that of the Macintosh.” The logic ofthe argument, both explicit and implicit, runs roughlyas follows. Miller [9] argued the capacity of short-termmemory (STM) is 7 ± 2. Exceeding the capacity ofshort-term memory degrades recall performance. Asnoted, the Macintosh has menus with more than 7items. Therefore, an interface such as the one found onthe Macintosh will be a bad interface because themenus exceed the capacity of short-term memory andwill lead to degraded human performance.

There are three problems that arise from applying thisguideline to this design choice. First, menu items aregenerally related, e.g. File, Edit, Format, or View.That is, the contents of a menu tend to form a singlechunk rather than a set of greater than 7 unrelateditems. Second, menu titles are generally mnemonicand so provide effective recall cues for the itemswithin. For both these reasons, menu titles aid ratherthan stress the user. Additionally, once the menu hasbeen pulled down, the contained elements are visibleand are now ‘knowledge in the world [5].’ None of theitems must be stored in working memory. UsingMiller’s work to proscribe the design of GUIs withmore than seven items per menu is misleading at bestand is analogous to claiming that a coin has a badinterface because people have trouble recalling all ofthe information on its face [10].

Although presented in the particular, this example isnot an isolated case. Several people have made this ora similar claim about the implications of Miller’s workfor human-computer interaction design. But what isimportant about this case is that it represents an attemptto get students to design from first principles. Thereason that this is difficult stems from the nature offirst principles and the resulting way in which they arestated. The statement “STM capacity = 7 ± 2” assumesbut does not state the range of cases to which it applies.Nowhere, is it made explicit that items can be eitheratoms or chunks; and that capacity is only an issuewhen external memory aids are not present. Thecontextualizing information that tells the designerwhen Miller’s work is appropriate is absent and so it ishard to use the abstract principle to either guide orconstrain prescriptions for action.

In other words, the designers of visualizations and thetools which are used to shape them need grounding inthe fundamentals of cognitive psychology in order tobe able to adequately explore the user's task orsituation for which the tool is being designed. They

need to be able to identify and to deal with two types ofpsychological knowledge. One type of knowledge isthat which will help the end-user solve the task relatedproblem, i.e., the problem of designing an appropriaterepresentation. The other type of psychologicalknowledge is that which enables the designer tounderstand the end-user's human cognitiveperformance strengths and limitations. It does littlegood and may even be harmful to create a visualizationwhich does not map in some relatively natural andcomfortable way to the cognitive structures of the end-user.

3. Concluding Remarks

In conclusion, designers and developers of tools andvisualizations need to have a strong understanding ofhuman-computer interaction and cognitive psychology.So long as it is easier to learn to write code than tolearn to communicate with your life partner it will beeasier to adopt and maintain a machine centered focusrather than a human-centered focus. Both knowledgeof the user's cognitive performance limitations andknowledge of how to design representations which willcontribute to the user's task relevant goals are essentialto the successful creation of new tools or visualizationswhich will be both useful and usable in dealing withvisualizations of abstract entities and relationships. Anunderstanding of basic HCI and human cognition willhelp designers and developers to remain focused on theproblems that should be solved rather than on theproblems which can be solved. Solving a technicallychallenging implementation problem may produce theimmediate gratification of having accomplishedsomething, but it represents a trap if that solution doesnot simultaneously contribute to the end user’s taskrelated goals.

Acknowledgements

The author would like to thank the anonymousreviewers for their comments on an earlier draft of thispaper. Work on this paper was supported in part byNSF CISE Grants CCR-9527130 and IIS-9873005.

References

[1] T. T. Hewett. Problem solving andcomputer graphics. Plenary speech atEuroGraphics ‘92, Cambridge, UK, 1992.

[2] ACM/IEEE-CS Joint Curriculum TaskForce. Computing Curricula 1991. TheAssociation for Computing Machinery, NewYork, 1991. (ACM Order No. 201880)

[3] T. T. Hewett, R. Baecker, S. K. Card, T.Carey, J. Gasen, M. Mantei, G. Perlman, G.

Strong and W. Verplank. ACM SIGCHICurricula for Human-Computer Interaction.The Association for Computing Machinery,New York, 1992 (ACM Order No. 608920)

[4] J. Preece, Y. Rogers, H. Sharp, D. Benyon,S. Holland and T. Carey. Human-ComputerInteraction. Addison-Wesley, Reading, MA,1994.

[5] D. A. Norman. Things That Make Us Smart.Addison-Wesley, Reading, MA, 1993.

[6] S. K. Card, J. D. Mackinlay and B.Shneiderman. Information visualization. InS. K. Card, J. D. Mackinlay & B.Shneiderman (Eds.), InformationVisualization: Using Vision to Think.Morgan Kaufmann, San Francisco, 1999.

[7] R. S. Palais. The visualization ofmathematics: Towards a mathematicalexploratorium. Unpublished manuscript

[8] A. Harrison, T. T. Hewett and R. K. Perline.Soliton viewer: a computational workbenchfor soliton geometry. Under review forpossible presentation at the UIST ‘99conference on user interface software toolsto be held in Asheville, NC, USA,November, 1999.

[9] G. A. Miller. The magical number seven,plus or minus two: Some limits on ourcapacity for processing information.Psychological Review. 63. pages 81-97,1956.

[10] R. S. Nickerson and M. J. Adams. Long-term memory for a common object.Cognitive Psychology. 11, pages 287-307,1979.

The Human-Computer Interfaces Course at IST

Mário Rui Gomes, Brisson Lopes and Manuel João FonsecaDepartment of Informatics Engineering, Instituto Superior Técnico

Lisboa, PortugalMrg@inesc.pt, brisson@ist.utl.pt, mjf@asterix.ist.utl.pt

Abstract

This paper presents the main and most relevantaspects of the Human-Computer Interfaces courselectured for the 4th academic years (and fifth year asan optional course) of the undergraduate program inInformatics and Computer Science at InstitutoSuperior Técnico, Technical University of Lisbon.The paper addresses the course contents and itspedagogical approach as well as the communicationfacilities used in the 1998/99 academic year tosupport the course operation.Keywords: Human computer interaction, teachingand learning.

1. Introduction

The main topic of a Human-Machine Interfacing(HCI) course is the design, development andevaluation of Interactive Computing Systems.Designing and developing human-machine interfaceshas become the most expensive factor in thedevelopment of information applications. Such costcorresponds today to roughly between 30 and 50%of the whole development cost1.However, even though the above is true, HCI hasbeen one of the less studied and less understooddisciplines2.The Design and development of HCI projectsrequires competencies and know-how in manyconcerning yet different areas such as SystemsEngineering, Programming Engineering, Graphicaldesign, Industrial Design, Psychology, Ergonomics,Project Methodologies, Interfacing Engineering(hardware), and Organisation Engineering, amongmany others3.These factors demonstrate how important it is theUniversities and Polytechnics include HCI Coursesin their curricula.

2. Background

In recent years, Portugal adopted the policy to fosterits national software industry. Many official bodieshave concurred to this with initiatives like the PEDIPProgramme, The Science Programme, The MosaicoInitiative and the Nónio Programme.

Many successful showcases in the development ofPortuguese products demonstrate the success of suchinitiatives. The Green Book on the InformationSociety produced by the Information SocietyMission of the Science and Technology Ministryextensively reports on the results.Software products must present the rightfunctionality and be easy to use in order to create astream of products and create and sustain a solidsoftware industry.In Portugal one can still find large competence gapsextending as far as the most successful products.This reason is more than enough to make the mainEngineering Colleges in the country to includecourses in HCI in their curricula. The softwareindustry has already acknowledged the need to focusintensively on HCI too as a leverage factor to staycompetitive in the international market.The importance of HCI goes beyond developing newsystems. In recent years, there has been remarkableraise in the use of integrated applications, such asSAP R3, that show a trend to dominate the market.Consultants and experts are required in order toparameterise and localise such integratedapplications so that they can be efficiently used. Thiswork includes both functional and the user interfacelevel. The later must be adapted to the actual endusers, the staff using the applications. In Portugal,there are some 60 enterprises that specialise inInteractive Computing Systems and recognise theimportance of the know-how on HCI. This spansnational borders and is acknowledged by bothnational and multinational enterprises involved ineither the development or localisation of InteractiveComputing Systems.In the United States, the National ScienceFoundation (NSF) sponsored a study on “NewDirections in Human-Computer Interaction,Research, and Practice". The report, published in1994, stresses the importance of HCI as one of themain areas in Information Engineering. However,the report acknowledges the many difficulties thathave prevented American universities from includingHCI courses in their curricula.The NSF proposes two ways to bring about thechanges it recommends in its report, which are:

• Integrate lectures on the design anddevelopment of HCI in the curricula ofInformation Engineering programs.

• Include HCI as one of the main components ofprojects assigned to students in the final (senior)year of undergraduate programs.

The report further recommends the acceptance ofHCI as a critical technology by universities andrecognises that the development of competencies andknow-how in HCI is a fundamental leverage factorin preparing students for their future professionalactivities in the scope of the Information Society.The above clearly stresses the importance of HCIfrom an integrated social and economic point ofview in those countries wishing to promote asustained Software Industry with a future.

3. Objectives

The definition of the objectives of an HCI coursemust be on par with market needs on skilledprofessionals. Therefore, students should aim at:1. Master the principles of information

presentation and user interaction, as well as,acquire the competence to evaluate the qualityand effectiveness of user interfaces.

2. Master the technologies to apply HCI principlesto products and the operation thosetechnologies's enabling tools.

3. Master the realisation of HCI projects in thescope of the most common environments (MSWindows, WWW).

The HCI course lectured at IST has a weekly load ofthree hours of lectures and two hours of laboratoryactivities. A project is assigned to the students at thefinal stages of the semester.The lecture time is dedicated to fulfil the firstobjective. The syllabus includes lecturing the basicprinciples that must be applied in order to achieveeffective information delivery to the user. Suchprinciples stem from mental models of informationprocessing and are a collection of methods andtechniques and good practice examples. These havebeen known and applied since long in areas likeadvertising.HCI derived its theoretical ground from a very largeset of experimental data collected over the years inthe laboratory and from field experiments. In thelaboratory session, students examine existinginterfaces and check their conformance against thetheoretical principles they learn. This activity isclosely linked with evaluation, either qualitative orquantitative (usability).The second aim of the HCI course (mastering HCIenabling technologies) targets at the analysis andproficient use of the technologies applied when

implementing efficient user interfaces, the evolutionof such technologies and their choice. In this, thecourse also aims at providing students with anoverview of the available technologies, theirevolution and capabilities to present information andinteract with the user.The third objective is reached by using a " learn bydoing" pedagogical approach with an HCI project.The project is carried out by students in the closingweeks of the semester. The project's main target is tolead students into applying the know-how they haveacquired and make it mature through actual use. Arather important point is the constant stress to be puton the pivotal role played by the end users in thedevelopment of the project. This is further enhancedby the project's requirement to conduct evaluationactivities at all the project's phases using real endusers for that effect.

4. Lectures

Lectures use the active discourse methodpedagogical approach with the support oftransparencies and examples of good (and bad)practice to demonstrate concepts.Students are submitted to five unannounced mini-testpapers in order to further ensure learning success andprovide students with feedback. These test paperstake place during the last 15 minutes of a lecture andaddress the lecture's topics.There is no sufficiently comprehensive textbookspanning all the subjects lectured in the HCI course.This makes it necessary to recommend severaltextbooks to students, with the latest edition of thetextbook by Shneiderman4 as a starting basis and thetextbooks by Preece5 and Mayhew6 covering thedetails in many areas. This dispersion stresses theneed for a new a more comprehensive textbook inHCI.

5. Laboratory Activities

The evolution of information systems design lead totoday's widely accepted client-server architecture.One of the best features of this architecture whenapplied to Interactive Systems is the clear separationof information processing from the user interface itprovides.On the other side, window management systems andrelated APIs evolved over the times. Recent yearshave seen the dominance of X/Motif (and itsderivatives) and MS Windows. Then the WorldWide Web made its first steps in 1994/95 andexperienced an exponential growth in the years thatfollowed. At the same time, the WWW increased itsinformation presentation and user interaction

capabilities due to the emergence of new orimproved standards of the HTML, Java andJavaScript languages. This, in turn, fostered the useof the enhanced capabilities by an ever-increasingnumber of services and users.The classes of services available extended to servicesto the citizen, education, entertainment, business,home banking, etc. Business services7 becameparticularly important and include such categories asadvertising, direct sales and employment offer. Ho8goes even further to propose three different businesscategories: product and services promotion, data andinformation services and business transactionservices.Because of the vast world opened up by the WWWhighlighted above, the WWW role in the InformationSociety and the WWW' programming and interactivepotential, the HCI targeted its laboratory and projectactivities at the WWW. These activities follow twomain guidelines:• Evaluation of WWW sites with special didactic

value, using case study methodologies, aimingat developing students' capacity for criticism andtheir proficiency in applying evaluation methodsand techniques to HCI design andimplementation.

• Exploration, discovery and application of theuser dialogue and information displayingpotential of the HTML and JavaScript languageswith the aim of promoting the skills studentsrequire to carry out their project assignments.

5.1. Evaluation of WWW sitesThe laboratory activities involving the evaluation ofWWW sites were carried out in three steps.In the first step, groups of students analysed andcriticised several proposed sites against thefollowing criteria:• Use of colour• Use of fonts and font sizes• Page layout• Look and feel• Navigation• Use of metaphorsThe second step had each group of students chooseone of the WWW sites and redesigning on light ofthe findings when applying the above criteria. Thenumber of student groups allowed more than onegroup to redesign the same WWW site. This wasimportant for the third step, as it will be shortlyevident.In the third step, each group of students presented itssite redesign proposal to all the colleagues attendingthat laboratory session. Each presentation includedthe group's reasons for the changes proposed. Aperiod for questions and answers followed the

presentation and all students were invited to criticisethe design proposals. It was extremely interesting towatch these proceedings since the degree ofinteractivity and commitment was very high andthere was the opportunity to address almost all therelevant aspects. Discussions among students lead tosynergetic results where several proposals werecombined into even better solutions.

5.2. Technologies (HTML and JavaScript)The laboratory sessions dedicated to HTML andJavaScript languages had the objective of promotingthe students’ competence in these languages, throughdemonstration and exploration of the languages’concepts and constructs. This would later on ease theuse of these languages when carrying out the HCIcourse project.The pedagogical approach to both HTML andJavaScript was the same. It started with thepresentation of the basic concepts and objects ofeach language. That was followed by languageconstructs of ever increasing complexity, alwayswith the support of illustrative examples. Studentswere asked to carry out some assignments at somecheckpoints. These assignments had the objective ofmaking students apply the knowledge they had justacquired and, at the same time, provide them withself-evaluation feedback.

5.3. Laboratory ResourcesLaboratory sessions took place in an electronicclassroom equipped with networked PCs withPentium II processors (300 MHz) and running underWindows NT. The network provided students withthe access to the WWW, file server services and on-line help facilities. The later included tutorial textsfull with interactive examples and on-line referencemanuals. The tutorial texts were also available inpacked format to allow students to take them homeand learn in off-line mode.

5.4. The HCI ProjectStudents were proposed several project themes tochoose from. Besides the obvious aim of carryingout an HCI project assignment, the themes proposedto the students were also aimed at developingstudents’ awareness of the societal environment forwhich they will be working after graduation sincethe themes spanned from products for disabledpersons to environment protection.

Theme A (Disabilities)This theme addresses the development of anElectronic Book to help parents, social workers andtherapists choose products for users with specialneeds and disabilities. Each Electronic Book had todescribe three products with similar functionality or

objectives and provide product comparison data andselection guidance. The products could be eithersoftware or hardware products.The actual product choice was left to the students ineach project team. Students had to select a type ofproduct and look for available products, collectingliterature and examples, if any. Many browsed theWWW for this. All products picked had to have beendesigned for users with special needs (elderlypersons, children) or disabilities (visual, hearing,motor, etc.). This requirement had the objective ofmaking students aware that HCI solutions have avast number of needs to address and are notrestricted to satisfying normal person needs (seeFigure 1).

Figure 1. Example of Project A.

Theme B (Environment Protection)This project theme addressed the development ofservices to the citizen to provide information on theenvironment, namely the quality of air in a region.Each project team had to select a type of end usersfrom the following three types:• general public• school children• environment expertsThe service had to be designed with the provision ofits deployment from information booths in mind. theservice had to provide information on air pollutantmeasurements of sulphur dioxide (SO2), ozone (O3),carbon monoxide (CO) and nitrogenous oxides(NO/NO2) concentrations in the environment. Theservice also had to provide appropriatedocumentation.The system had also to allow its users to select thedata to be displayed, i.e., from the current day, thelast week or the last month. Automatic datacollection by the system was simulated with a data

file containing all needed pollutant concentrationvalues to ease system construction (see Figure 2).

Figure 2. Example of Project B

Others projectsSome students proposed other type of projects someof them were accepted. The most relevant was thenew site of the Informatics infrastructure thatsupports all the courses on Informatics, ElectricalEngineering and Informatics Management, the RNL(Rede das Novas Licenciaturas) (see Figure 3).

Figure 3. RNL Site

6. Communication Between Students andProfessors

One of the subjects addressed by any HCI coursemust be Computer Assisted Person to PersonCommunication. Therefore it made all the sense toaddress this subject with very special care anddemonstrate its usage and benefits when carrying outcommunication between students and the course’steaching staff. This systematic pedagogical

approach made all students to have intimate contactwith different technologies used today for Person toPerson Communication and realise the advantagesand disadvantages of each of these technologies andwhen and how to use them.In the HCI course the following three Person toPerson solutions were used:• electronic mail• newsgroups• a WWW siteThe HCI course WWW site aimed at makingavailable to students the most complete and up todate information on the HCI course itself. At thebeginning the site contained only information on thecourse’s syllabus, its teaching staff (with e-mail andpersonal WWW pages addresses), how the finalmark would be calculated and important dates anddeadlines.As the semester went by other information wasadded like:• marks achieved by students in test papers• themes for the course’s project• tutorial sessions support documentsThe objectives of the newsgroup that was set-upwere to publicise general-purpose announcementsand putting and answering general questions,including questions on parts of the course contentsthe students might not have understood. Thenewsgroup aimed at providing students with themeans to anticipate the answers to students’recurring questions. Students were advised to perusethrough the newsgroup before coming forward witha question or remark so that they could see whetherthat subject had already been put and answered.One remarkable result was the speed at whichstudents knew of their marks on test papers. Usually,these are posted at assigned places on the walls ofthe administrative area. The HCI course followedthis rule but, at the same time, posted the marks onits WWW site and announced the fact through thenewsgroup. The result was that many students cameto know of their marks well before the administrativestaff had time to post them at the usual places.Moreover, the number of students looking for theirmarks in the later place was very small.Finally, electronic mail was used only when thequestions and matters to attend to were of a privatenature and for internal communication within theteaching staff.

7. Conclusion and Final Remarks

The experience gained in the 1998/99 academic yearallows us to say that the HCI course at IST wassuccessful since most of the objectives aimed at wereachieved. However, there is still a long way to go

before an HCI course with a stable syllabus is inplace and that that course answers all learningobjectives set. One important point to be addressedin the future is the publication of adequate andcomprehensive literature and books. The other pointis the creation and use of more interactive means fordemonstration and exploration of the course’scontents.

8. Acknowledgements

The authors want express their deepestacknowledgements to Professor Lourenço Fernandeswho, at the time the undergraduate program inInformatics and Computer Science was created atIST, had the perception of the necessity and thevision to create a course with the multidisciplinarycharacteristics the HCI course has.

References

[1] B. A. Myers, M. B. Rosson, "Survey on userinterface programming". In: Bauersfeld, P.;Bennett, J. & Lynch, G. (Eds.) CHI´92Conference Proceedings: ACM Conferenceon Human Factors in Computing Systems.Addison-Wesley; pp. 195-202 (1992).

[2] G. W. Strong, "Introductory Course inHuman Computer Interaction”, ACMSIGCHI Bulletin, 20 (3), pp. 19-21. (1989).

[3] R. M. Baecker, et al, “Readings in Human –Computer Interaction: Towards the Year2000”, Morgan Kaufmann, (1995).

[4] B. Shneiderman, “Designing the UserInterface – Strategies for Efective Human –Computer Interaction”, 3rd ed., AddisonWesley Longman (1998).

[5] J. Preece, et al, “Human – ComputerInteraction”, Addison Wesley Longman(1994).

[6] J. D. Mayhew, “Principles and Guidelines inSoftware User Interface Design”, Prentice-Hall (1992).

[7] J. H. Armenix, "Close Readings" of InternetCorporate Financial Reporting: Towards aMore Critical Pedagogy on the InformationHighway, The Internet and HigherEducation, vol. 1, no. 2, pp 87-112 (1998).

[8] J. K. Ho, “Evaluating the World Wide Web:A Global Study of Commercial Sites”, J.Computer Mediated Communication, vol. 3,no. 1 (1997),(http://207.201.161.120/jcmc/vol3/issue1/ho.html).

Media Education at the Bauhaus-University Weimar

Charles A. Wuthrich

CoGVis/MMC�

Department of MediaBauhaus University Weimar

99421 Weimar, Germanycaw@medien.uni-weimar.de

Abstract

The rapid evolution of Media through the introduction of digitaltechnologies, the multimedialization of computer hardware and thedevelopment of new network technologies have generated the needfor new profession profiles responding to the rapid changes in thisfast evolving branch. In the new Media, arts and humanities aretied to rapid technological development to create new markets andnew products. This paper describes the study courses taught at theDepartment of Media of the Bauhaus-University in Weimar, wherefuture Media Designers, Media Critics and Media Scientists areforged. All courses are taught in a project based way, so as to teachflexibility for a rapidly evolving sector. The new study courses rep-resent a first attempt of definition of an education at the leadingedge of technology which is entirely based on graphics, multime-dia and network-based paradigms.

1 Introduction

In recent times, the information society is transforming from anindividual or organization based archive (or repository) of informa-tion to be processed to a networked environment allowing commu-nication between humans and within society [7]. Older broadcast-based media, which allowed essentially a one-way communication,have been completed by Internet technology, which instead allows atwo way communication on the same medium. The development ofcomputer networks, as well as the introduction of powerful graphicshardware on computers, have allowed the introduction of computer-based communication: computers have evolved from data process-ing units to hardware for a new medium allowing a new type ofinterhuman communication bringing in a new quality, and allow-ing multimodal location-independent communication. New power-ful hardware allows the generation of shared virtual environments,where communication can happen between virtual representationsof users (avatars), allowing the decoupling of the representation ofthe subject from his physical appearance.

The establishment of these new technologies requires new spe-cialized study courses which differ from classical Engineering,Computer Science, Artistic and Human Sciences education [8, 4],integrating different traditional fields [5, 2]. The market requirescreative and innovative graduates capable of coping with a rapidlychanging industry and with the challenges of three-dimensionalrepresentation [1]. Moreover, the scientific foundations for suchsystems are still at their infant stage, there is no general model, andsystems are developed on the stimulus of application, and are notgeneralizable.

The establishment of a Department of Media at the Bauhaus-University in Weimar and the introduction of two study courses in“Design of Media” and in “Media and Cultural Studies” in 1996

�Computer Graphics, Visualization, Man-Machine CommunicationGroup

have awaken great interest in the German Media Community. Onone side, for the first time in a German University a Departmentwas established at the crossroad between the Humanities, the Artsand the Sciences with the purpose of forming a generation capableof researching, developing, designing and analyzing products in afast evolving medialized world which is being created through theintroduction of new graphical, acoustical, multimedia and commu-nication technologies.

The Humanistic-Economic department component is repre-sented by the presence of professorships in “Perception Theory,Theory and History of Media”, “ Theory and History of Artifi-cal Worlds”, “ Sociology” and “Media Management”, while De-sign professorships have been established for “Multimedia Story-telling”,“ Design of Media Environments”, “ Media Events”, “ Trendsand Public Images”,“ Interface Design”, “ Experimental Radio” and“Electronic Music Composition”. Instead of following the classi-cal single medium oriented education, the design professorshipshave been conceived to have designers capable to utilize more thanone medium for their expression. The technological component iscurrently represented by professorships in “Databases and Com-munication Systems”, “ Networked Media” and “Computer Graph-ics”, and will expand next to include professorships in “VirtualReality Systems”, “ Computer Supported Cooperative Work” and“Computer-Based Gaming”.

The Faculty of Media understands itself first of all as a thinktank and as a confrontation location. A think tank where the crit-ics of existing and of new forms of media serve as a stimulus forthe conception and for the development of new media, both as newforms of representation with traditional media as well as new typesof technologies. The mission of the department is to study and crit-icize existing media, to research and develop new interhuman com-munication paradigms, and to promote the creative use of existingmedia.

The focus of the two study courses currently available is mainlyset on the Arts and Humanities, whereby students have the possi-bility of receiving a strong education on the available technologiesso as to be able to follow and to participate to their development ascompetent initiators and co-concievers. This approach is motivatedby the observation that many of the technologies which were de-veloped in the last thirty years in this field could not make a break-through in the market due to the lack of applications based uponthem. The closeness of end users to the developers of these tech-nologies is a key to the success of the technologies themselves. Endusers should be able to have a constructive exchange with the devel-opers and conceivers, so as to both influence the development andbe aware of it in time.

To complete its offer, the Department plans to introduce a thirdstudy course in 1999 on Media Systems, so as to insure the presenceat the faculty also of technology developers aware of user needs.The new study course aims at the market of the rapidly expandingmedia, multimedia and telecommunication industry which createstechnologies and systems for the new media.

All the study courses are organized in a project-oriented way:students do not learn a theory that in future after the diploma will beapplied in practice. Instead, the study courses are organized aroundthe “learning-by-doing” principle. This is motivated by the speedof the development of technology: the goal of studying should behere to train flexibility and the development of new paradigms,more than building the knowledge for a lifelong profession. At avery early stage, students are therefore involved in the faculty re-search, development and artistic activities and participate in the de-velopment, accompanying the work on the projects with traditionalcourses relevant for the projects themselves. Furthermore, basicsare taught through introductory courses which have to be completedbefore gaining access to the faculty projects.

This paper is structured as follows: section 2 presents the currentarts and humanities study courses in Media, section 3 presents thenew study course in Media Systems which completes the offer ofthe department. In the last section, perspectives for graduates aretraced and some conclusions are drawn.

2 Technological Education and Multime-dia for the Arts and Humanities

In the education of artists and of human scientists working withthe media, a twofold goal has to be accomplished: on one hand,traditional media, such as film, broadcasting, print media have tobe taught; on the other hand, students have to be prepared to the useof new media for their projects, and to successfully cooperate to thedevelopment of new technology. The contribution of artists to thetechnological development is essential, since end user feed back isof extreme importance.

Since the media industry develops itself at the same time towardsmultinational media corporations and into tiny services deliverers,a strong component of economical foundations have to be taught.This is achieved through the teaching of basic economical coursesas well as special courses on the economics of startup companies.

Both study courses are centered on projects. Under individual tu-toring, groups of students are assigned a theme each semester whichthey have to develop. Project results are then publicly presented atthe end of the semester. This way of teaching focuses the study onthe production process, from conception to execution.

Artists and Humanists are known to suffer of the typical com-puter device fear. Moreover, most students approach the studycourse attracted by the contents of media, not by the technologydriving them. However, in a high-tech environment, it is very dif-ficult to avoid being confronted with the glitches and problems ofthe technology. This is why a comprehensive technological compo-nent, including three basic courses on the foundations of Mathemat-ics, Computer Science and Media Systems (network and broadcasttechnology) are taught. Additionally, fundamentals of program-ming languages, of operating systems, of communication networks,of computer graphics, of multimedia, of databases, of image pro-cessing, of sampling algorithms and of interactive technologies aretaught on a regular basis. Students are also confronted with thebasics of hardware development.

They are allowed to learn broad and comprehensive foundationsof the technologies they will use during their lives. Teaching is notapplication-oriented or software oriented as in most institutions, butconcentrates on principles, so as to avoid an application-specificapproach, which has no sustainability on the long term. This ulti-mately trains towards flexibility. Additionally, students are allowedto visit introductory courses on the main interaction, animation,multimedia, modeling, rendering, image processing, compositing,sound editing and web based techniques so as to understand theprinciples of tool based work in the media. This is done throughhands-on intensive tutorials aiming at real production.

The aim of this approach is to promote the competence in stu-dents so as to allow them to dialogue with other people involvedin the media industry, whatever their personal inclination is, be ittowards the arts, technology or on reflection on the media, and topromote the understanding of the basic technologies available.

The job perspectives of the students are improved consider-ably by the in depth technological education which improves theirchances in the work market. The traditional gap between researchand development, and end users is minimized. Arts and humani-ties students cultivate in the non-technological study courses (whichstill constitute the main part of their study) their characteristics asartists and critics, and pursue careers in the creative or humanisticfield, but are able to understand the potential development in tech-nology and to accompany it from its very early stages. This is adecisive advantage with respect to traditionally schooled studentsin the arts and humanities.

3 Media Systems: a New Paradigm forMultimedia Education

Media Systems are systems that have an influence on the Media,i.e. that support and allow communication between people. Thisincludes broadcast, networked and point to point communication.Setting the accent on systems underlines that focus is set on com-plex entities, rather than individual hardware components thereof.Media Systems are therefore complex systems used to communi-cate.

The study course is addressed therefore to all persons that wantto acquire technical/scientific knowledge in the new media, andwant to pursue a technical career in Digital Media, Gaming andEntertainment, Computer Graphics, User Interfaces, Animation,Image Processing, Visualization, Networks, Computer SupportedCooperative Work, Teleactivities, Virtual Reality and Digital andMultimedia Publication, and that want to participate to the tech-nical conception and to the development of the expanding mul-timedia, entertainment and communication industries. The studycourse builds ultimately professionals that are creatively capableof conceiving systems that revolutionize the way of thinking andof communicating between humans, and that favour new forms ofexchange between them. Furthermore, since the study course is in-tegrated in a faculty where a wide variety of study profiles for themedia industries are present at the same time, it prepares throughdialogue and discussion with the creative and humanities facultycomponents to cooperation with all professionals of the media in-dustry and research.

The study course global duration is four years, subdivided in twomajor slices of two years each.

3.1 The propedeutic years of the study course

The rapid developments in the multimedia and telecommunica-tion industry require a total rethinking about how a scientific studycourse should be organized. Technology evolution is so fast, andnew application areas and environments develop so quickly, that itis hard to predict what an optimal study curriculum is for the field.What are needed both in industry and in science are flexible personsthat are capable of working in projects with minimal introductiontimes, and capable of working as a task force on a theme for a pe-riod of time in an innovative way, then move to new projects.

Instead of the traditional approach based on trying to cover dur-ing the study the whole spectrum of related matters until “universalknowledge” of the matter of study is reached, the study course istherefore organized around the concept of research project, and triesto teach “high-tech” on a learning-by-doing fashion. Of course, it isnot possible to renounce completely to introductory courses, which

provide the tools for scientific reasoning. However, introductorycourses should be kept to a minimum, so as to leave as much roomas possible for project-based work.

From the point of view of content, the study course is centeredon the idea of a system. Basic courses are taught during the first twoyears. They serve three main purposes: first they provide the math-ematics and computer science basics for mathematical reasoningand for programming, second they provide an overview of systemmodeling which is relevant for the field and which is available inscience. Third, the study of available media technology, includingtraditional media, is provided as a foundation. The modeling ofcomplex systems requires the mastering of all possible modelingmethods, which are then taught in the basic courses.

More in detail, the Media Systems introductory courses are con-stituted by the subjects listed in Table 1. Purpose of these introduc-tory courses is to provide the student with the capability of abstract-ing and understanding complex systems and their components. Thenumbers to the right of the tables indicate the number of semesterweekly hours of course during the study course, which have to bemultiplied by 1.5 to obtain the credit points in the ECTS1 givenby the single units. Computer Science and Mathematics rudimentsare built through the courses listed in Table 2. All the above men-tioned courses have to be absolved in the first two years of the studycourse.

Note the centering of the subjects on the concept of systems andof mathematical modeling, and the relatively light Computer Sci-ence and Information Theory contents with respect to traditionalInformatics and Electrical Engineering. The faculty of Media of theBauhaus-University strongly distinguishes these two subjects fromMedia Systems, the latter involving more analysis and modeling ca-pabilities than the first two, this in order to master the complexityof the systems to develop.

In order to prepare the students to the last two years, where themain part of the time will be spent in research projects, students areallowed to choose during the second year two labs where they canparticipate to the department’s research activities in two chunks of 8hours per semester (in study projects) so as to gradually familiarizewith the research activities. Here, students have to bring their con-tribution to research, starting from the entry level of developmentand get acquainted with the methodology of study of the secondpart of the study course.

3.2 The second part

In the second part of the study course students have to work onthe department research projects. Each semester, the student has towork in one of the laboratories of the faculty at a research projectwhich is assigned to him at the beginning of semester. This meansthat students will have the possibility of working on four researchprojects in the last part of his study course, for a total of 16 hoursper week presence at the lab each semester, which gives them 24credit points if the project is concluded and defended successfully.

The department is currently setting up laboratories centeredaround the research themes listed in Table 3. Additionally to thework on projects, students have to get credits for accompanyingcourses in the quantity of 4 weekly hours per semester, giving thema total of 6 ECTS credit points per semester. The courses can bechosen from the advanced courses list, which are variable offers ofthe faculty, depending on the projects available, but which mightinclude the course catalog listed in Table 4 The aim of extensivework on the research projects, and of the theoretical foundationswhich are delivered through the courses, is to train the students tobe able to take up a research project having only basics knowledge

1ECTS stands for the European Credit Transfer System, which is theofficial exchange system among European Universities and provides trans-ferable credits to exchange students.

and acquire during the development of the project the necessaryfoundations on the project themes so as to deliver research results.

In the course of their study, students enter their first projects (al-ready in the second year) as executors, and acquire more and moreexperience, creativity and dialectic capabilities as they becomemore mature and gain broader knowledge, participating more andmore to the conceptional phase and to bigger parts of the projects,until they are able with in the final thesis to take a project and de-velop it by themselves. This of course requires extensive individualtutoring, which has to be performed directly by faculty. In return,faculty is allowed to concentrate most of the time on its researchthemes and can profit from the work of the students. Through anadmission exam and through a personal interview of the applicantsselection of suitable students for such a non-conventional studycourse is achieved. The individual nature of the study course re-quires not only application, but also particular creativity of thought.

4 Perspectives and Conclusions

According to the last German Federal Ministry of Economics re-port on the Information Society [3] the average rate of growth ofthe entertainment, telecommunication and multimedia industry inthe early nineties has been over 20% yearly. This growth rate hasbeen rapidly increasing since, and creates an immense demand onspecialized personnel, in all sectors, be they in the creative, orga-nizational or technological sector. Graduates from flexible studycourses at the leading edge of research stand excellent chances forconducting an interesting work life. Moreover, their flexibility “pre-programs” them for continuous education, in a world where it is lessand less possible any more to learn at the university a professionwhich will last a whole life.

With its three study courses centered on research, on the appli-cations and on the critics of Multimedia and Network technology,the Department of Media at the Bauhaus-University in Weimar isan unique example of cooperation among people originating fromtraditionally distinct academic paths. This reflects in the educa-tion, which centers, of course, on the special skills required fromscientists, artists and humanists, but completes them with elementswhich are necessary to ensure the capability of all three componentsto dialogue with the other participants to the technology develop-ment and usage. Furthermore, with its transdisciplinary approach,the department follows the path traced by the interdisciplinary re-search group at Xerox Palo Alto Research Center (PARC) in theearly seventies [6] which developed the mouse-based desktop userinterface, adapting the approach to respond to the needs of a broadersubject.

The applications for access to the new study courses state theinterest of youth for the perspective careers involved, and are cur-rently beyond 1000% the number of admissions available in allstudy courses currently available.

Acknowledgments

Special thanks are due to the people who contributed with their supportand discussions to the contents of this paper: Lorenz Engell, Peter Hupfer,Gerd Zimmermann, Grit Th¨urmer, Claus Pias, Bernd Schalbe, and G¨untherSchatter. A special thank is also due to all functionaries at the Ministry ofResearch, Science, and Culture of the Free State of Thuringia who supportedour Department at various stages of its development.

References

[1] E. A. Abott. Flatland. Dover Publications, Toronto, Ont., 1992.

[2] J. D. Foley. The convergence of graphics and imaging.Com-puter Graphics Forum, 17(3), 1998. Keynote Speech at Euro-graphics ’98.

[3] J. M. Harms. Computertechnik, Telekommunikation, Unter-haltungselektronik und Medien wachsen zusammen. InDie In-formationsgesellschaft: Fakten, Analysen, Trends. Bundesmin-isterium fur Wirtschaft, Bonn, 1995.

[4] W. A. Ross and M. W. McK. Bannatyne. Creating a 3-dimensional paradigm as a foundation for graphics programsin higher education. InWSCG’99, Proceedings of the SeventhInternational Conference on Computer Graphics, Visualizationand Interactive Digital Media ’99, pages 510–516. Universityof West Bohemia, Plzen, Czech Republic, 1999.

[5] Stanford University. Image system engineering.http://www-ise.stanford.edu .

[6] C. P. Thacker, E. M. Craig, B. W. Mapson, R. Sproull, andD. R. Boggs. Alto: A personal computer. In D. Siewiorek,G. Bell, and A. M. Newel, editors,Computer Structures: Read-ings and Examples. McGraw-Hill, New York, N.Y., second edi-tion, 1981.

[7] D. Tsichritzis. The dynamics of innovation. InBeyondComputation, ACM 50th Year Anniversary Book. CopernicusSpringer Verlag, 1997.

[8] D. Tsichritzis. Reengineering the university.Communicationsof the ACM, 1999. In print.

Media Technology Broadcast and Network Technology, 8Communication processes and standards.

Psychophysiological Systems Perception Theory, Human Senses, Psycho-physiology4Physical Systems Mechanics, Dynamics, Electromagnetism, Waves 6Logics and Discrete Systems Logics, Semantics, Discrete Mathematics 4Concurrent Systems Concurrent and Distributed Systems and Processes 4Fuzzy Systems Analog and Fuzzy Systems, 4

Non-quantifiable Systems (Quality, Expressivity)Stochastical Systems Statistics, Probability theory, Empirical Methods 4Projecting Methods Theory and Methods of Design 4Cultural and Social Systems Sensemaking Systems, Symbol Systems 6

Narration, Social Systems

Table 1: Media Systems introductory courses

Introduction to Computer Science 4 Higher Mathematics 3Device Architecture 2 Numerical Mathematics 3Algorithms and Data Structures 2 Discrete Mathematics 3User Interfaces 2 Functional Analysis 2Operating Systems 2

Table 2: Computer Science and Mathematics courses

Title Description

Tomorrow’s Realities Virtual Reality, Computer graphics, Computer-augmented reality,Real-time systems, Visualization, Digital sound

Learning and Teaching Teaching at distance, Tele-teaching, Multimedialization of courses,Continuous Education

Storytelling Narrative techniques, Esthetical processes, Sensemaking systemsInformation presentation, Computer Animation

Entertainment Games, Computer supported Gaming, Media-based EntertainmentWork Computer Supported Cooperative Work (CSCW), Tele-working

Work organization of tomorrow, TelematicsInterfaces 3D-Interfaces and -I/O, Haptic Displays, Immersion,

Human perception

Table 3: Research themes for the department laboratories

Game Theory Geometry Computer Graphics MultimediaStorytelling Image processing Image Recognition Global IlluminationTypography Virtual Reality Digital Television Sound ProcessingColour Theory CSCW Learning at distance TelematicsEthics and Media Empirical research Visualization AnimationElectronic Communication Electronic publishing Interaction Creative ConceptionErgonomy and work science Database Management Operations Research Media Cultural

Systems Studies

Table 4: Optional courses for the second part of the study course

Defining Interactive Multimedia Design Education:Expanding the Boundaries

Bonnie MitchellBowling Green State University

Bowling Green, Ohio, USAbonniem@creativity.bgsu.edu

Abstract

The future of interactive multimedia design educationdepends upon the disintegration of discipline-basedboundaries within the higher education system.Disciplines outside of computer science andcomputer art contribute significantly to theunderstanding of the cultural and social implicationsof the medium. A cross-disciplinary approachenables students to explore new ideas that informtheir understanding of the implications of multimediaproduction on the individual and society. Theevolution of communication design has reinforced theneed for an understanding of human behavior andrepresentation. Because functionality no longer takesprecedence over appearance in WWW and CD-ROMdesign, students must have a strong understanding ofthe principles of art and design as well asprogramming. The boundary between contentprovider, visual designer, and software designer isblurring therefore reinforcing the need to explore therelationship between art, psychology, computerscience, popular culture, software engineering, andthe social sciences in a meaningful way and integratethese connections into the curriculum. Only thenwill higher education begin offering interactivemultimedia design education that will enablestudents to influence the direction computer graphicsand interactive techniques in new and profound ways. Keywords: curriculum, interactive multimediadesign, cross-disciplinary.

1. Introduction

In the past decade, academic institutions have foundthemselves inadequately prepared to cope with therapid changes brought about by the cultural appeal ofcomputer graphics and interactive multimedia. Tenyears ago, who would have thought that URLswould be printed on cereal boxes, and teenagerswould prefer online chat sessions to TV? A lifeenriched by video games, special effects in film, andthe World Wide Web often entices students to seekout academic institutions that are able to preparethem to become interactive communication designersand/or technicians. To provide a well-roundededucation in interactive multimedia design,discipline-based degrees should be rethought and

new multi-discipline fields developed. This ideatranscends the common practice of cross-disciplinarycurriculum requirements and focuses on thedevelopment of hybrid courses and initiation ofacademic de-categorization. The development of anew interdisciplinary structure could enable flexibleadaptation thus encompass the changing needs ofstudents entering the interactive multimedia designfield.

2. Collapsing the Structure

We are currently immersed in the technological andideological transformation of global cultures. Notonly is information on virtually any topic availableto anyone with access to a networked computer, butthe hierarchical structure controlling informationproduction and publication is collapsing. In a worldwhere scholarly texts and student productions co-exist side-by-side, we find that our expectations ofand experiences with information acquisition havechanged. According to Margot Lovejoy, “In theInternet, we have the commingling within a vastglobal community of international resources of bothhigh and low cultural appeal. Some see the screen asa linguistic leveling device which flattens anddevalues language. Others see a rich evolution oflanguage occurring on the Web, not as it evolves innovels but as an experience of communicationbeyond the usual hierarchies of mediatedpublishing.” [1] Not surprisingly, students often relyon the Web for information and we have seen URLsreplace print references in their research papers. Theyrealize the value of easy access to this cornucopia ofdiverse information and want to be involved in theproduction of information as well a consumer of it.

3. Revisiting the Past

Along with the collapse of the informationproduction hierarchical structure, the method andmeans of conveying information have begun to takedivergent paths. Whereas we see a resurgence of thedependence upon visual communication, the meansof communication has evolved in a direction thatdefies the foundation upon which communicationwas founded. The earliest tangible examples ofcommunication were often found on cave walls in the

form of depictive images. In these ancient culturesinformation was exchanged directly by means ofvisual, auditory and gesture-based reciprocation.This interactive interchange was confined to thephysical locale. Over the years, cultures havebecome less and less dependent on images and haverelied on the abstraction of symbols such as text toconvey thought. With the advent of the book,information became tied to a tangible material thatenabled the dissemination of ideas to a wideraudience. Books became permanent records ofknowledge and were often considered the authority.Cultures developed elaborate campaigns to maturatetext-based literacy, while de-emphasizing the visualimage. Interaction, inherent in pre-text basedinformation exchange, was severed, leaving thereceiver of information alone without recourse. In“Being On Line: Net Subjectivity”, DoctressNeutopia states “There is something dead about theword. After it is written, like a photograph, it isthere as a reflection of past thought, there to remindus of a moment in time when the inspirationpossessed the writer”[2]. The transient nature of theWorld Wide Web suggests there is no authority andinformation is constantly in flux and very muchalive. Knowledge becomes a process of explorationand evolution, not a product.

4. Crossing the Boundaries

The re-emergence of an emphasis on visual,interactive exchange of information in the late 20thcentury generates the need to expand ourinterpretation of what it means to communicate. TheInternet has become a fertile ground for experimentalforms of social exchange. The enabling of immediatereciprocation that transcends geographic spacechallenges the formative structure of exchange yetbroadens the appeal. Abraham, Meehan, andSamual, in the book “The Future Ain’t What itUsed to Be”, stated “Online culture is, more thananything else, about satisfying our fundamental needsfor dialogue, discourse, and debate”[3]. In the past,personal interaction within local communitiesnourished the culture. Today the interchange withinon-line culture supplements and often surpasses localexchange enabling not only dialogue and debate butalso cross-cultural transmutation of ideas.Understanding the visual image plays an importantrole as students begin to understand culturaldifferences and explore new modes of exchange.

5. Education’s Responsibility

With the responsibility of preparing today’s youth tofunction effectively within a cross cultural,interactive, visual environment, educators have anobligation to address the concerns of the changing

state of information consumption and production. Inmany countries, we are seeing new academicprograms emerge and existing structures expand toemphasize the production of visual information usingcomputer graphics. The technical aspects involvedin the production of images and interaction usingcomputer technology typically consume academicattention. Little, if any, time is devoted to howthese images are transforming our culture. Theessence of interaction and information design isoverlooked in favor of the technical means ofproduction. We teach programming which focus onsoftware development and art classes where studentsdesign Web pages, but where do we teach coursesthat integrate the mathematical, artistic, and culturalaspects of new media and communication design? Across-disciplinary approach to interactive multimediadesign education will enable students to produceworks that not only stimulate our senses, but alsochallenge us intellectually.

6. The More We Know, the More WeCan Say

Whether we design interfaces that empower orentertain, we are attempting to create screens thatconvey information to the user. Visualcommunication, whether employed in scientificvisualization or in art, is often more successful whenknowledge of human perception is applied.Developing an understanding of the structure andfunction of an image will enhance the student’sability to communicate the essence of a concepteffectively. One of the strengths of computer graphicsis its ability to render obscure and abstracted data. Itis able to do so only through detailed analysis of thedata that makes up the image. Knowledge is power.As stated in the classic computer science proverb,garbage in – garbage out, the quality of the inputheavily influences the quality of the output.Students with a deeper and broader base ofknowledge are able to produce more sophisticatedand engaging interactive computer graphic products.

7. Communicating Ideas

In the Computer Art curriculum at Bowling GreenState University students were given a JavaScriptassignment to develop WWW pages thatinvestigated the concept of time as it relates tohuman perception (see Figure 1). The idea wasdiscussed as a class, thus generating initialapproaches to the subject. Sub-topics were identifiedand students selected which theme they wanted toresearch further (see Figure 2). Students were giventime to search the Internet, go to the library, findimages, and think about a specific approach theywanted to take with the project. The following

week, the students met with a small group of otherstudents that selected a similar theme and eachstudent described the specific idea they wanted toconvey visually and interactively in their JavaScriptproject. Often, the topic involved abstract contentsuch as Mat Brandeberry’s JavaScript project thatexplored the idea of duplicity, doldrums, anddecisions (see Figure 3). Rudolph Arnheim in “Artand Visual Perception” states that “The relationshipbetween intellectual knowledge and visualrepresentation is frequently misunderstood. Sometheorists talk as though an abstract concept could bedirectly rendered in a picture; others deny thattheoretical knowledge can do anything but disturb apictorial conception” [4]. Students grappled with theidea of visually communicating these abstract ideaswithin an interactive environment. They alsoconsidered how interaction could effectively enhancethe concept they were trying to convey. Thetechnical aspect of how to develop an interactiveexploration of their idea using JavaScript was nothighlighted until the research was well underway andthe idea was solidified.

Figure 1. This is the start of the JavaScript Project.

Figure 2. The topic was subdivided into themes.

Figure 3. This approach explored the concept ofduplicity, doldrums, and decisions.

8. Interface Design - Human Factors

As the students moved from idea to digital imageproduction to interface development, issues related tohuman factors and effective visual layout arose.According to Brenda Laurel, “The workingdefinition the interface has settled down to arelatively simple one - how humans and computersinteract - but it avoids the central issue of what thisall means in terms of reality and representation” [5].Students needed to decide whether an exploratory orintuitive interface would be more effective incommunicating their idea. They also discussed howthe experience they were providing for the viewertranslated into meaning. If a student decided toexplore the concept of time, technology, and

intimidation, an interfaced designed to confuse andbelittle the user would communicate more than anycollection of images. Yet, if the student wished toempower the user, an interface that translatedtechnology into human terms would result in a morerelevant experience. The design of the interaction,interface, and visual communication should beseamlessly interwoven into the expression of theidea. Interdisciplinary educational experiencesprepare students to make meaningful connectionsbetween disparate fields of study.

9. Interaction Production Education, orTechnology is Never Enough

One of the obstacles that prevent the development ofan integrated interactive multimedia designcurriculum is that it does not fit nicely into anycurrently defined field. In an essay entitled, “TheDesign of Interaction”, Terry Winograd states,“Although computers are at the center of interactiondesign, it is not a subfield of computer science... Aswell as being distinct from engineering, interactiondesign does not fit into any of the existing designfields. If software were something that the computeruser just looked at, rather than operated, traditionalvisual design would be at the center of softwaredesign. If spaces were actually physical, rather thanvirtual, then traditional product and architecturaldesign would suffice”[6]. Often we see academicprograms such as Computer Art, Computer Science,Information Technology, and Public Communicationoffering classes in interactive multimedia, each withits own unique approach to the subject. Sinceinterface and interaction design relies heavily ontechnical proficiency, artistic talent, and anunderstanding of human behavior, all aspects of thefield are rarely explored in any depth in any singlediscipline. Nicholas Negroponte believes, that“...we are moving away from a hard-line mode ofteaching, which has catered primarily to compulsiveserialist children, toward one that is more porous anddraws no clear lines between art and science or rightbrain and left” [7]. At the university level, thismethod of teaching is slow in developing.Occasionally computer art professors teach basicprogramming and the mathematical terminologynecessary to understand 3D modeling, but it is rareto find courses in computer science or informationtechnology that also focus on the principles of art anddesign. It is even more rare to find interactivemultimedia courses that address issues of humanperception, behavioral science, global studies, andpopular culture.

10. Technical Proficiency

Within the Computer Art classroom, the studentslearning JavaScript to convey their ideas inrelationship to time were forced to rely on a smallsubset of the programming language with very littleknowledge of the concept of object-orientedprogramming. Because time was devoted to critiqueof the artwork and interface design as well asdiscussion, very little time was left to investigate thelanguage. Without prerequisite knowledge ofprogramming languages, the art students werestumped by concepts as fundamental as variables.Ten different JavaScript examples were presented andthoroughly discussed. These examples were madeavailable to the students via the Internet. Needlessto say, very few student projects code thatsignificantly differed from the example codedistributed. Did they understand the code? Yes, toa very limited extent. Did they understand thepotential of the language and how to manipulate it tosolve unique problems? No. As a beginning levelassignment, the project was successful in revealingthe artistic potential using computer graphics,interface design, interaction, and JavaScript, but tofully understand the medium, more programmingexperiences would be necessary. The students alsowere not given the opportunity to study the impacttheir work had on the global viewer nor how it fitinto popular culture.

11. A New Generation

After completing the project, the three studentvolunteers developed the interface that linked theindividual projects together under one unifiedstructure. The project was published on the WWW.Although this was a student project, it was availablefor the world to peruse. For many students, this wasthe first time they had published a work of art. It isimportant to note that the average university studenthas a very different relationship to technology thantheir professors. Most of the students in theComputer Art course were between the age of 19 and21. On the average, they were 5 years old when themicrocomputer became widely available. Most hadgrown up with computers in their home. They donot remember a time when video games were notpopular. Fax machines, photocopiers, and answeringmachines have always existed according to theirrecollection. They were in middle school when theWorld Wide Web hit the mainstream. Email, chatrooms, and URLs are as much a part of their lives asTV was a part of ours. The immediacy of real-timeinteraction with remote parties via the Internet is notnew to this generation; it is an integral part of theirlife. Yet, communication that transcends culturalboundaries still has its novel appeal. As educators

we need to be aware that as consumers of informationon the Internet, our students are often quite adept.Yet as producers of information, they are often veryinexperienced.

12. Global Communication

Although publishing interactive information thatreaches a global audience is exciting for the student,the lack of physicality, and the cross-culturalimplications should be addressed in the curriculum.Creating interactive work that accepts or monitorsresponses from global visitors enables the student tobegin to understand how the work reads in a cross-cultural virtual world. Oftentimes a student’s visionof the world is based on their personal experiences.Lectures, books, and TV may present alternativeviews, but reality still exists within individualempiricism. Using the Internet to initiatecommunication and solicit responses to studentinteractive multimedia works, enable the student tobegin to understand the implications andinterpretations of their work. The World WideWeb enables the student to build pages and publishtheir work. By also including feedback forms on thepage, the work not only becomes public but also issubject to global critique. A narrow, uninformedview of a topic or a design that does not complimentthe content will quickly have its weaknesses revealedby the global community. It is often the case thatcritiques come from many different cultures and frompeople with various professional interests. Within asingle discipline based course, this global feedbackenables the student to integrate ideas that transcendgeographic and discipline-based boundaries.

13. New Solutions

Within the academic community we are searching fora home for Internet-based interactive multimediawithin existing departments and schools. Yet thevery nature of the medium is cross-disciplinary. Tobe successful, an educational experience in interactivemultimedia should span across academic boundariesand possibly include sub-topics within computerscience, computer graphics or computer art,traditional art, information science, social science,psychology, business, and philosophy. Studentsalso need to be more aware of the global economyand cultural similarities and differences. Currently,in most universities, students must select a major inone of the above areas and hope to take courses in theother areas. Contemporary solutions to providing awell-rounded experience in interactive multimediahave often fallen into four categories: new programs,experimental courses, advising, and prerequisitecourses. Academic institutions are often reluctant toestablish new programs therefore faculty initiatives

are essential to the educational success of studentsfocusing in this area. Creative solutions come fromthe realization that new media has changed the waywe, as a culture, process information. Broadeningour own experiences to encompass new multi-disciplinary approaches to our own understanding ofthe issues will inevitably translate into attainablesolutions.

14. The Team Teaching Dilemma

One of the answers to the problem is to provideteam-taught cross-disciplinary courses that integratedesign, programming, and human factors as well asinformation from other supporting fields. Fornumerous reasons, this solution is often unattainable.Although many universities pay lip service to thebenefits of interdiscipline studies and team-teaching,faculty are often discouraged by the lack of support interms of course load reductions, salary compensation,and access to information and facilities. Facultyeager to cross boundaries often find themselveswithout the support of the university and unable toinitiate new course development withoutcompensation and approval. Supportivedepartments often find that budgetary structures arenot flexible enough to compensate two professorsfully when together they teach only one course.Also, it is common that team taught courses requiremore preparation because of the ongoing negotiationinvolved between professors. Despite theseobstacles, faculty occasionally sacrifice summers andadequate pay to develop experimental courses andintegrate new cross-disciplinary content into theirinteractive multimedia courses. These heroic effortshave not gone unnoticed. The introduction of multi-disciplinary study, complimented by multipleviewpoints, increases the students understanding ofthe medium in which they work.

15. New Approaches

Although team teaching enhances the educationalexperience of both the student and the faculty, thediscipline-based boundaries within the universitystill prevail. Art and Computer Science departmentsoperate independent of one another and proudlyeducate students by delivering the content they knowbest. Courses that do not fit the idealistic vision ofwhat art or computer science should be are oftenshort lived or never approved. Many universitieshave developed departments or programs that attemptto merge the various interests involved incommunication technology education. With titlessuch as Multimedia Technology, InformationScience Technology, and Visual CommunicationTechnology one would think that an aspiringinteractive multimedia producer would be able to

obtain a well rounded education in the field. This issometimes true, yet not always. These newdisciplines also tend to become exclusive rather thaninclusive and course content follows a prescribeddirection that may or may not include the multi-faceted field of interactive multimedia development.Imagine the possibility of developing a departmentthat employed art and design faculty, computerscientists, popular culture researchers, architects,social science faculty and experts from othersupporting fields. The goal would be to makeconnections between the various disciplines and tie itall together using computer graphics and interactivemultimedia development as the catalyst. This cross-disciplinary approach to education would posechallenges as the various fields seek out commonground and merge interests. The rewards wouldreveal themselves not only in the educationaloutcomes of the students but also in the intellectualgrowth of the faculty as they broaden theirexperiences and modify their single disciplineapproach to education. In an actively evolvingcurriculum such as this, it is likely that the studentswould not only be the recipient of ideas but alsoengage in the process of creating the connectionsbetween disciplines.

16. Conclusion

As more and more students enter academicinstitutions with the desire to work with interactivemultimedia, the demand for a cross-disciplinaryapproach will increase. The preparation necessary toenter the field will necessitate new disciplines orpossibly a destruction of the academic walls thatprevent students from focusing in a multi-disciplinary direction. There will be more demandfor a team teaching approach and a broad-basedcurriculum. Students have been, and will continueto challenge existing degree requirements toaccommodate their desire to integrate computerscience, art and other supporting areas. Thedevelopment of a diverse educational approachpromises to reveal exciting new opportunities notonly for students but also for faculty and theinstitutions themselves. The knowledge andexperience that comes from multiple viewpoints andvarious fields of study will challenge the students tobecome independent and critical thinkers. Studentsinvolved in a fully integrated cross-disciplinaryinteractive multimedia design educational experiencewill gain the understanding and power to takeinteractive multimedia to new exciting directions.This new generation of communication engineers andartists will inevitably alter the course of thisimportant medium thus forcing academic institutionsto again re-assess and refine their approach toproviding interactive multimedia education.

References

[1] Lovejoy, Margot, Postmodern Currents: Artand Artists in the Age of Electronic Media,Second Edition, Prentice Hall, New Jersey,1997.

[2] Avillez, Martim, Being On Line: NetSubjectivity, Lusitania Press, New York,1996.

[3] Abrahamson, Vickie, Meehan, Mary,Samual, Larry, The Future Ain’t What itUsed to Be, Riverhead Books, New York,1997.

[4] Arnheim, Rudolph, Art and VisualPerception: A Psychology of the CreativeEye, The New Version, University ofCalifornia Press, Berkeley, 1974.

[5] Laurel, Brenda, “ Computers as Theater”,Addison-Wesley Publishing, New York,1993.

[6] Winograd, Terry, “The Design ofInteraction”, Beyond Calculation: The NextFifty Years of Computing, Copernicus,Springer-Verlag, ACM, New York, 1997.

[7] Negropone, Nicholas, Being Digital,Vintage Books, New York, 1995.

MODULARITY: TH E KEY TO COURSEWARE RE-USE

José Carlos TeixeiraDep. Matemática – Fac. Ciências e Tecnologia da Univ. Coimbra – Portugal

teixeira@mat.uc.pt

ABSTRACT

By courseware modularity we mean the possibility to build complete courses from manysmall course modules fitting together. In contrast to puzzle pieces, course modules should bedesigned to fit into many different situations, allowing different courses to be produced.This goal introduces a lot of problems to be solved, like organisation (hierarchical) of thematerial, formats’ compatibility, clear separation between contents, structure andpresentation, …

To allow an effective multiple-use of courseware (use of the same courseware entities bymany users), as well as its re-use (use of the same courseware modules in the production ofdifferent courseware entities), the courseware production should be performed in smalllearning units that keep courseware highly modular and self-contained. Problems with thedidactic design should not omitted.

The adaptability (meeting individual learning characteristics and learning needs) is also areason to support the modular approach. Nevertheless, development is needed to ensure, in apracticable way, the adaptability of courseware.

The storage and management of the learning material is also an issue. The use of distributeddatabase systems and the integration and inter-operation of several services, allowing theaccess to the best fitting learning material for the user and learning target, have to beachieved to allow a real multiple use and re-use of learning material.

Last, but not the least, the production cost: the courseware production is expensive andrequires large teams of persons with different skills and expertise, and it is very time andresource consuming. One way to lower production costs without affecting the material itself,could be the material re-use.

COURSEWARE RE USE OR COURSEWARE TO REUSE?

José Carlos Teixeira♣, César Páris♦, Joaquim Madeira♣, J. Brisson Lopes♠♣ Dep. Matemática – Fac. Ciências e Tecnologia da Univ. Coimbra - Portugal

♦ Centro de Computação Gráfica – Coimbra - Portugal♠ Instituto Superior Técnico – Lisboa – Portugal

teixeira@ccg.uc.pt; cparis@ccg.uc.pt; jmadeira@mat.uc.pt; brisson@ist.utl.pt

1. Introduction

Teaching and training scenarios have changedover the years. This does not come as a surprise,but one must study what are the correspondingimplications in current teaching and learningmethods and processes.There are three major differences. First, starterlearners are more receptive to learning if they areable to visualize different objects, models andconcepts they are growing up in a TV andcomputer world , and by actively manipulatingthem (learning is thus a hands-on process,which stimulates the curiosity of learners and theirinterest in acquiring additional knowledge.Furthermore, there has been a huge advance intechnological support (computers and networks),and, finally, learning is no more confined to thefirst 25 years of life, but has turned into a lifelongprocess.Today's learning scenarios extend beyond thetraditional classrooms: students can accesscourseware materials from home and teachers feelthe need to use new media in their classes. Bothscenarios require multimedia material and addinteractivity to the traditional chalk and boardscenario.In this position paper we want to open thediscussion about the main issues that must beaddressed to enable courseware reuse and thefeasibility of possible solutions, based on someexperiences we already have

2. Production of Courseware Material

Multimedia courseware production is not a one-person job, but requires large teams of personswith different skills and expertise, and is very timeand resource consuming. There are two ways oflowering costs: increase the number of peopleusing each course, or make use of the material indifferent contexts, that is, reuse it.The best way of guaranteeing that the coursewareproduced fits its author’s idea is that he/she willproduce it all. In fact, no deviations will be madeconcerning the focus and original design, and the

author will always be up to date with the currentstate of production. With some tools to keep trackof the courseware produced he/she will easilymaintain and validate it. However, one canimmediately perceive the drawbacks of thisoption: time constrains, tools and methods usedmight not be the best, but always the ones that theauthor can access. The best option will be then tohave separate teams, according to specificfunctions.

Idea Generation

Content Selection and Strucuture

•Source selection•Evaluation•Migration•New material•References -> Nodes

Didactic Review

Assessments Tests

MM Production

Programming Simulations

Design and LayoutCreation

Quality CertificationCycle

Delivery

Management Process

Fe

edb

ack

by

the

re

vie

wer

s

Se

lf te

sts

give

n b

y th

e S

tude

nts

Figure 1. The courseware productionprocess.

As depicted in the above diagram, the same personcan perform different functions. For example ateacher, or a group of teachers, can generate oneidea, put it on paper and take the responsibility ofthe didactic and assessment aspects, leaving themost technical aspects - MM Production,

Programming Simulation and Design, as well asLayout Design - to experts in those fields. Ofcourse the authors and the members of theproduction team will have to interact, and that canbe done in real time or in deferred time. This lastoption could be very helpful if thinking globally,i.e., if a scenario is planed where authors arescattered throughout the world and the productionteam fixed in one specific place.The drawback of this scenario is that there aremany extra variables that are input into theprocess. There is now the need for a ManagementProcess ensuring that there is a correct workflow.Furthermore, since the authors are not alwaysdirectly involved in the production, there must becontrol mechanisms allowing them to validate thesuccessive production steps. Nevertheless, at theend of the process there should also be a QualityCertification Cycle, which guarantees that thematerial to be released is still valid.Although this additional production structure, isassociated with an increase in overall costs, it willnow be possible to generate high-qualitycourseware with short-production cycles - two keyfactors when it comes to courseware production.One way to lower production costs withoutaffecting the structure that produces them, couldbe the re-use of material already produced. Thenumbers in the IDEALS project [1], an Europeanproject with 13 partners, from industries touniversities, from 6 different countries, show thatit is possible to lower production costs making re-use of existing courseware.In that project the conclusions showed that theproduction time for just one study hour was 164.3hours [2]. This ratio depends too much on the typeof media involved (pictures, video, sound) and thedemanded quality. Using templates for some partsof the course the number of hours required toproduce one hour of course material was loweredto 91.8 h, with savings of about 44%. That is onlythe amount of time associated with coursewarecreation and does not include the time consumedby management tasks.More important than the above is the possibility tofurther reduce the production time of coursematerials by re-using previously developedmaterial. This allows saving about 27 workinghours per one hour of courseware. However, it wasobvious to all partners involved in the field tests,that re-using implied almost always additionaladaptation times which were strongly related to thecomplexity of the material being re-used. Theaverage time for this was about 6.0 hours per onehour of courseware produced [3]. All of those

tasks took more time than expected, mainlybecause there were no real, effective guidelines.The numbers presented above just give an ideaabout the effectiveness that can be achieved by re-using courseware. It is obvious that the results willvary depending on the quality of courseware andthe level of re-use that it is planned - just media, orchunks of knowledge.General-purpose courseware will lower productioncosts and can be distributed to a larger number ofpeople. On the other hand, knowledge is gettingmore and more specific today, and coursewaremust follow this tendency. Besides developingcourseware from the scratch, one needs to be ableto adapt general-purpose courseware to specificgroups.Note that generic software should be highlymodular to ensure its use by large, heterogeneousgroups of users, and has to be designed andproduced with great pedagogic and didactic care,as well as carefully designed screen layout anduser interface. Thus making it easy to use, as wellas stimulating, and ensuring its success regardingthe learners. The cost associated with ensuring ahigh-degree of courseware quality is usually offsetby the number of learners that can use suchgeneric courseware.On the other hand, specific courseware can easilyincorporate higher-level knowledge units andmight be less modular. This reduced degree ofmodularity and higher specificity usually restrictthe number of possible learners, and the usual highdesign and production costs might not be as easilyoffset by the actual number of learners.

3. Courseware Reuse

Nowadays we see images being reused (e.g.,clipart), but this is not enough. There should be away for teachers to reuse compact units ofknowledge (modules), which focus on a specificsubject. It will be necessary to build coursesbearing one word in mind: modularity. Whencontent is stored apart from sequence andstructure, one can start thinking in reusing thestructure itself.Three hierarchical levels that can be defined whendiscussing reuse: the most atomic level - pictures,movies; modules -, a collection of atomic entitiespresented at the same time to a learner, andlearning sequences - structures referencing themodules and their relations.The easiest and obvious thing to do is reusingatomic entities: you just need to replace onepicture for another or one movie for the other.There is no secret here: if you have a database

with some indexation you can ease your work. Interms of presentation, maybe you need to do someadjustments/conversions, but in fact this is not acritical issue.When you want to re-use modules the firstproblem appears: you must present these modulesto an audience. So if you want to re-use them, youmust have coherence, and for that the use oftemplates is inevitable. Having different templatesbuilt according to guidelines will make thispossible. This factor is indeed the most criticalone. There must be tools and processes that caneasily map atomic objects into modules accordingto the template elected.Finally the learning sequences, i.e., a collection ofreferences to modules and to other learningsequences that together build one knowledge unit.It is important that these learning sequences fullydescribe all knowledge that has to be presentedregarding specific subjects, allowing their re-use indifferent contexts, without any dependencies toother modules or sequences. Once more, this mustbe stated in the design guidelines and followed byall authors.If you are thinking about a book, the modules willbe the sub-chapters and the learning sequenceswill define which and how the sub-chaptersconstitute one chapter. For the whole book, thereis a learning sequence that references otherlearning sequences (chapters).Courseware developers must pay close attention indeveloping an appropriate set of guidelines forcourseware design. Those must be as rigid aspossible, in order to achieve global coherence forthe course material to be produced. Changes to theguidelines will generate new course material thatmight not be compatible with that which has beenpreviously developed. When this situationhappens, probably previous courseware must bere-produced, and then you start wondering if re-use is effective or not.Based on the many experiences of re-usingcourseware, and despite the level of re-usabilitythat is in question, it is known that authoringguidelines cannot remain fixed, or at least withoutsignificant changes, for a long period. The same iscommonly accepted as necessary to produce agood and stable Resource Bank of courseware,which will allow effective cost reduction inCourseware Production. So, what is the answer?The method of developing a courseware reusablelibrary must take into account that the guidelinesmight change, and then treat all coursewarecomponents as entities that need to be stored andclassified in an effective way. The same goes forthe learning sequences, modules and presentation

templates. All of what is stored in the ResourceBank can be modified/replaced individuallywithout affecting others.We can easily foresee teachers having availableseveral course templates, choosing the ones whichare most suitable for their specific courses, and,after that, filling in the empty spaces withpreviously produced modules. If no suitablemodules can be found, then they will producethem and make them available.The courseware developer must structure itscourse, looking up what he can reuse from theResource Bank, and create or modify the missingentities, using the guidelines pre-defined. Then thecourseware information entities are accessed andintegrated inside the presentation template chooseand accordingly with the learning strategy to befollowed. Courseware modules are integrated, likea jigsaw puzzle, in order to fulfil coursewareauthors’ needs. It is not expected that they createtheir own courseware just reusing previouslyproduced modules, but rather that they willreorganize it, adapt some parts and create others.Each information system that is worthwhile hasthe same problem: it grows unchecked until it istoo hard to manage. An approach to support theprevious ideas in a community of producersdemands that it must be possible to create andmaintain a Knowledge Centre.It must also be ensured that everybody usescompatible technologies, and it must beguaranteed that there are common rules,procedures and standards. The use of databases isinevitable, due to the expected large amounts ofinformation produced. These databases mustreflect the following key issues: cross-platformaccess, modularity, scalability and, of course,different access permissions. Classification andindexing are other important problems that mustbe addressed.Multimedia courseware production requiresmultidisciplinary production teams, whilepublication requires clear business models.Copyright ownership has to be transferred, butauthor rights remain with the creators. This showsthe need for adequate business models.Furthermore, courseware delivery releasesmaterials that can be copied over and over againwith no quality loss because they are in digitalformat. This issue can only be solved withappropriate and secure copyright marking tools, aswell as commonly accepted copyright policies.

4. Conclusion

This position paper raise some thought inspiringmain issues that have to be addressed whenconsidering the design and production re-usablecourseware modules.The authors hope that these ideas might contributeto fruitful discussions during the workshop, aswell as possible future chances for collaborationbetween workshop participants.

References

[1] IDEALS – Integration of DEDICATED forAdvanced Training Linked to Small andMedium Enterprises and Institutes ofHigher Education, Project Programme EUTelematics Applications Programme,Project nr. ET-1012, 1995,(http://ideals.zgdv.de)

[2] IDEALS, Report on the Restricted DemoCourse. IDEALS – project of the EUTelematics Applications Programme,Project nr. ET-1012, 1998,(http://ideals.zgdv.de)

[3] Paaso, J, Computer Based TeachingTechnology for Software EngineeringEducation, Acta Universitatis OuluensisC123, 1998, Department of ElectricalEngineering, Univeristy of Oulu, Finland

Framework for Deve lopment of Course Material for Teaching Computer Graphics

V. V. KamatDepartment of Computer Science & Technology,

Goa University, Goa 403206, Indiavvkamat@unigoa.ernet.in

Abstract

In this paper we define the framework fordeveloping course material to teach ComputerGraphics courses. The paper will mainly discusswhat support material is desired for effectiveteaching/learning the subject rather thanimplementation of a teaching tool. It lays emphasison three important components of the courseware,mainly the design of course content, anexperimenter’s workbench and an assessment tool.The framework envisages the need for anintegrated approach to courseware developmentfor teaching / learning Computer Graphics. Thisapproach would clearly separate the role ofcourseware developer from that of the deliveryperson (traditional teacher). The main theme ofthe paper is to develop learner-centric coursecontent that will depend less on the teacher andmore on the learner. Such a courseware wouldsupport three independent perspectives. Firstly,the developer’s or author’s perspective, whichwould view the course content as data or mediaobjects that could be easily shared, maintainedand reused in other courses. Secondly, thelearner’s perspective, which would accord thelearner a platform-independent experimentalworkbench to learn by hands-on experience.Finally, the administrator’s perspective whichwould collect feedback on each learner's progressand performance.

Keywords: Web-based education, Instructionaldesign, Computer graphics courseware.

1. Introduction

Computer Graphics as a discipline has maturedover the last three decades. Today manyuniversities teach more than one course onComputer Graphics at different levels to studentswith background in science, engineering and finearts. Often the objective and the pre-requisites foreach of these courses are different although thereis fair amount of overlap in its content. It is

visualized that good course material for teachingthese courses would contain high volume ofmultimedia content such as visuals, animations,audio, video and simulations besides the text.Therefore to design the course material forcomputer graphics course has always been achallenge to both the educationist and thetechnologist. Kayssi et al. have discussed similarissues in a recent publication on the application ofweb-based tutoring as applied to computernetworks courses [1].

There is no doubt that for an effective delivery ofthe course, a lot of efforts go into the preparationof course material to be used in the class room bythe instructor. The additional support materialsuch as student notes, laboratory assignments,reading material and self-tests, further enhancesthe quality of instructions. For a highly developednetworked infrastructure, which is available today,it is unfortunate that each instructor has to developthis material from scratch. It is an accepted factthat good experienced teachers are difficult tocome by and this approach will have limitedapplicability and will not scale-up.

One important aspect of any course on ComputerGraphics is experimentation. Typically theinstructor gives a series of assignments andstudents solve these assignments and learn theconcept by actual implementation. The commondifficulty faced by the student duringimplementation is that they have to build theapplication from scratch. This can be very timeconsuming and frustrating exercise.

It is also noticed that students who likeprogramming otherwise are some timesintimidated with matrix notation, vector algebra,3D geometry and Calculus. Often these subjectsare taught in the first few years of collegeeducation and by the time student opts forcomputer graphics course, these concepts areforgotten. As a result, students associate computergraphics to mathematics and develop anunnecessary fear for the subject. This makes it

mandatory for the instructor to review some ofthese mathematical preliminaries in the class,before proceeding to teach Computer Graphicsconcepts and algorithms. Further, the approachestaken to teach a fundamental course and anadvance course may be different. For example, in afundamental course on Computer Graphics, moreemphasis may be laid on data structures andalgorithms and a student may be expected toimplement these concepts from ground-zero. Onthe other hand, for an advance course, theemphasis may be more towards making use ofcanned code in order to build reasonably complexvisualization programs.

2. Graphics Courseware Components

In order to create resource material to be sharedamong a group of instructors and the students,adequate measures should be taken at the designand development stage. It is a matter of fact thatthe teachers often prefer to improve upon theirown course material after each delivery of thecourse. Therefore, it is important that such acourse material is highly modular in content,reusable, easy to maintain, store and distribute.Further it should provide the scope for theinstructor to customize it to keep the learnermotivated.

For quite some time now, a need has been felt bythe Computer Graphics community to define aGraphics Workbench, a tool to carry outexperimentation in visualization. Such aworkbench would allow the students to experimentwith Computer Graphics concepts without havingto write a lot of additional code. It would requireto have easy to learn easy to use API that wouldallow anyone to implement a graphics module thatcan be designed at a higher level, independent ofany specific platform and tested on a modestsystem with requirements commonly available atwork and home.

Assessment is an essential part of any training. Anassessment tool should not only collect thestatistics about usage, evaluation and progress of aparticular learner but also provide assistance to theauthor to prepare questions and answers for an on-line test.

The framework suggests that all three componentsare essential to give an integrated view of thecourseware to the learner.

3. Instructional Design Strategy

The development of course content is an extremelylabor intensive and time-consuming process.Therefore it is important that proper care is takenat the instructional design stage so that the coursecontent can be easily modified and maintained.One method to do this is to design course materialin a modular way with a hierarchical structure ofconcepts and separation of course structure fromindividual concepts [2]. Each concept will be anindependent entity in itself with well-definedobjectives, pre-requisites and evaluation criteria.Course material organized in this fashion can bemodified easily without much effort. There can bemore than one structure to organize thecourseware. This will allow two learners withdifferent learning profiles and taking the samecourse, to be presented with different learningmaterial. Further, it is observed that if the studentsdo pre-work before arriving at the training session,the learning process then becomes proactive ratherthan reactive resulting in better comprehension andretention.

Important processes in courseware design wouldinvolve the definition of a set of concepts. Eachconcept would define the following:

• Concept objectives• Any pre-requisites• Relation to other concepts• Questions & answers• Testing material• Trainer’s guide• Student’s material (pre-work)• Classroom slides• Lab Assignments & Solution• Reading material

The courseware design team would involve a setof subject experts, an instructional designer, acouple of programmers, and graphic artists. It isvery important to realize that the development ofgood instructional material lies much beyond theexpertise of subject experts. It definitely involves ateam effort and cross-disciplinary approach. Therequired skills could be best compared to that ofweb page design if not to making a film. Sinceseveral subject experts could be involved, atemplate has to be prepared in order to provide auniform structure for developing course content.Such a template can also help in identifying theproperties of the content, which can be used laterfor managing the content. [3]

4. Architecture

The suggested architecture for development ofcourseware is based on client/server model andwould use a relational database as a centralrepository for storing different media objects. Thevast popularity of the Internet and the World WideWeb has made the Web browser as the mostconsistent cross-platform solution for browsingmultimedia data. This would also provide acommon user interface to developers, learners andadministrators, a platform independent client fromany location on the Internet/Intranet. Theinteractivity can be achieved through cross-platform languages such as Java. For example, tocommunicate the concept of interactive curvedesign, a Java applet can be written to input thepoints interactively and drawing a Bézier curveand then displacing one of the points andredrawing the curve. Any amount of hand wavingin a classroom cannot communicate this concepteffectively but a simple demo will do the trick.

5. Graphics Workbench

The experimentation Graphics Workbench (GW)will basically have a visual experimentationenvironment that can allow a learner to easily setup and test customized sequences of a graphicspipeline. It can not only help the learner to createand edit complex geometrical models but also testones own module by plugging it along with othermodules pipelined in a complex data flow graphcalled module network. The workbench willperform automatic module scheduling in a modulenetwork. This will allow the learner to modifyparameter for each module in the network,automatically re-scheduling all modules affectedby a parameter value change. It will support richset of graphic file formats and have large set ofgraphics modules arranged in a hierarchical libraryfor the fast and task oriented retrieval.

Typically a computer graphics application can bemodeled as a set of cooperating processes of threetypes as shown in Figure 1.

• Input processes, responsible inputting thegeometric model into the system

• Computational processes responsible forperforming one or more graphic operations onthe geometric model

• Presentation processes and display processes,responsible for presenting and storing theoutput result.

When designing a stand-alone application, onetypically would describe the application in a top-down fashion. The description would begin withthe overall structure of the application and wouldend with a detailed description of each processincluding its inputs, outputs and function. This canhelp the learner to rapidly prototype an applicationusing the GW.

Figure 1. Data flow graph showing anetwork.

6. Evaluation Tool

This tool would help the teacher and the learner totrack the progress and performance of anindividual. In addition, it would keep a statisticsabout the difficulty of the question by its usage. Itwill consist of set of forms and wizards that wouldallow the author to create variety of objectivequestions of the type:

• True and false questions• Multiple choice questions• Match the columns• Questions using interactive simulations

Each question and its answer are stored in thedatabase to form a question bank. Each time astudent wants to take the test, it would select thequestion randomly form the question bank, set thenumber of questions and the time. The wronganswer to any of the question would link back toconcepts in the courseware. It may also include

3D Data 3D Data

Transform

Projection

Clippin

Display Save

3D Data 3D Data

Transform

Projection

Clippin

Display Save

hints, instructions and advise to a particularquestion using different media types such as text,audio or a video.

7. Conclusions

The concept paper outlines the need, objectivesand suggests an integrated approach to coursewaredevelopment for teaching / learning computergraphics. It envisages that such a coursewarewould be able to support three independentperspectives each for the author, learner and theadministrator so that effective learning can bemonitored closely.

References

[1] Kayssi A., El-Haji A., El Assir, M.,Sayyid, R., Web-based tutoring andtesting in a computer networks course,Computer Applications in EngineeringEducation Vol. 7 No.1, 1999

[2] Teixeira J. C., Paris C., Brisson LopesJ.M., Paiva, A.M., Authoring Coursewarefor Distributed Environments ProceedingsICVC’99 held in Goa, India, February1999

[3] Connolly C. Developing Learner CentricContent, White paper EmpowerCorporation, 1998 (http://www.empower-co.com)

List of Participants

• Anand, Vera – Clemson University – USA

• Bailey, Dan – UMBC – USA

• Bailey, Michael J. – SDSC – USA

• Bresenham, Jack – Winthrop University – USA

• Brodlie, Ken – University of Leeds – UK

• Brown, Judith R. – University of Iowa – USA

• Case, Colleen – Schoolcraft College – USA

• Chiappini, Giampolo – Consiglio Nazionale delle Ricierche – Istituto per la Matematica Applicata – Italy

• Cunningham, Steve – California State University Stanislaus – USA

• Dochev, Danail – Institut of Mathematics and Informatics, Bulgarian Academy of Science – Bulgary

• Eber, Dena E. – Bowling Green State University – USA

• Ebert, David S. – Univesity of Maryland – USA

• Fuchs, Laurent – IRCOM-SIC, Université de Poitiers – France

• Gomes, Mário Rui – IST/INESC – Portugal

• Hansmann, Werner – University Hamburg – Germany

• Hewett, Thomas T. – Department of Psychology, Drexel University – USA

• Hitchner, Lewis E. – California Polytechnic State University – USA

• Howard, Toby – University of Manchester – UK

• Kamat, Venkatesh – Goa University – India

• Kjelldall, Lars– KTH – Sweden

• Leitão, J. Miguel – Instituto Superior de Engenharia do Porto – Portugal

• Longson, Tony – California State University – USA

• Lopes, João Brisson – Instituto Superior Técnico – Portugal

• Madeira, Joaquim – Universidade de Coimbra – Portugal

• McGrath, Michael – Colorado School of Mines – USA

• Mitchell, Bonnie – Bowling Green State University – USA

• Morie, Jacquelyn F. – Rhythm and Hue Studios – USA

• Mumford, Anne – Loughborough University – UK

• Ollila, Mark – University of Gavle – Sweden

• Owen, G. Scott – Georgia State University – USA

• Páris, César – Centro de Computação Gráfica – Portugal

• Pavlov, Radoslav – Institut of Mathematics and Informatics, Bulgarian Academy of Science – Bulgary

• Prieto, Andres I. – University of Cantabria – Spain

• Santos, Beatriz Sousa – Dept. Electronica e Telec., Universidade de Aveiro – Portugal

• Santos, Manuel Próspero dos – Departamento de Informática da Universidade Nova de Lisboa – Portugal

• Slavik, Pavel – Czech Technical University – Czech Republic

• Sowizral, Henry – Sun Microsystems Inc. – USA

• Teixeira, José Carlos – Universidade de Coimbra / Centro de Computação Gráfica – Portugal

• Tomida, Akemi – University of Cantabria – Spain

• Trindade, Jorge F. – IPG-ESTG – Portugal

• Wolfe, Rosalee – DePaul University – USA

• Wüthrich, Charles – Bauhaus - University of Weimar – Germany