Upload
yuriy-sountsov
View
223
Download
1
Embed Size (px)
DESCRIPTION
The time is nigh for architects to seize the tools that await them...on THE GRID.
Citation preview
T H EG R I D
Acknowledgments
Development of THE GRID would not have progressed without
the support of Art Lubetz, Josh Bard, Dale Clifford, Kai Gutschow,
Johnathan Kline, and Ramesh Krishnamurti. Also, many thanks
go to Sam Sanders, Adam Lans, Mike Jeffers, Ben Finch, Yeliz
Karadayi, Claire He, Paulina Reyes, Alex Fischer, and Emerson
Stoldt. Last but not at all least, the support of my parents was
invaluable.
Thesis Website - the--grid.tumblr.com
Secondary Thesis Website - whatisthegrid.tumblr.com
CMU SoArch Thesis Website - www.andrew.cmu.edu/course/48-509/
CMU SoArch Home - www.cmu.edu/architecture/
Contact:
Yuriy Sountsov
Find me on LinkedIn! Revision 10 - 5/8/20141
Fig. 0.1 QR code for the thesis website.
Fig. 0.2 QR code for the tutorials.
Acknowledgments and Contacts 1
Table of Contents 2
Part One: THE GRID Compiles 6
Part Two: THE GRID Executes 74
Part Three: THE GRID Terminates 144
Ta b l e o f Co n T e n T s
T H e G R I D
2
5
6
Table of Contents 6
Introduction 7
Interest and Architecture Moving Forward 7
Advisors and Primary Contacts 9
Project Brief 12
Project Methods and Timeline 17
Research 21
Precedents 21
Literary Research 25
Interviews and Reviews 37
Software Research 43
Hardware Research 55
Deliverables 57
Applications 57
Moving Forward - Software Package 61
Moving Forward - Benefits and Death 63
Moving Forward - Imagination and Experience 69
Pa R T on e - Ta b l e o f Co n T e n T s
R e a l T I m e 3 D V I s u a l I z a T I o n
7
Fig. 1.1 The eye provides the most powerful sense humans have: vision. Architecture is often a primarily visual profession - while many architects argue the tactile and auditory aspects of architecture are also very important, the experience always comes back to the appearance of a building. Therefore it should be of paramount importance to architects how they communicate the visuality of their designs, yet one of the most powerful tools in an architect’s arsenal, the computer, remains wholly unused.
I, Yuriy Sountsov, was interested in this project because I had
the opportunity to give something to the field of architecture that
it has struggled to have. During my last year at Carnegie Mellon
University I had the time, resources, and commitment necessary
to put forth a complete, developed, and forward-thinking project
that others could take and use in their lives as designers and
practitioners of architectural theory and thought.
In the five years that I spent studying architecture at
Carnegie Mellon University...I had seen the future. And it was a
strange future, indeed. The world, reader, was on the brink of
new and terrifying possibilities. But what was made available in
my education was severely lacking. Architects spend too long
learning tools that are obsolete by the time they find ways to
teach those tools to new architects.
What if the world could see inside the mind of the architect?
What if the architect’s ideas did not travel a maze before
becoming visible?
Architects are ready to learn. One of the major aspects of
an architectural thinker is that they are open to new ideas, new
societies of thought. Over the centuries, it has taken radical
I n T R o D u C T I o nIn T e R e s T a n D
aR C H I T e C T u R e mo V I n G fo R w a R D
8
thinking of the likes of Brunelleschi, Gaudi, and Candela to
advance the field of architecture in great leaps and bounds, but
it was not because they created things that had never been seen
before but that they knew what was available and created what
could be possible. The digital world is only the latest arena which
is thus untapped. It has been exponentially growing for decades
and the time is nigh for architects to seize the tools that await
them...on THE GRID.
Fig. 1.2 Brunelleschi’s dome, a single combination of previously disparate concepts that allowed architecture to take a great leap forward.
9
Yuriy Sountsov - Yuriy Sountsov is a fifth year architecture
student at Carnegie Mellon University. He is dissatisfied with
the digital backwardness of the program he has been exposed
to and wonders sometimes whether architects have become so
desensitized to the creative world around them that they think
they are on the cutting edge when in fact they are on the cutting
block. He has experience with various digital design software,
various video game engines, has seen many films and has explored
film technology. He sees a problem in architectural practice and
wishes to contribute his time and energy for free to fix it.
Arthur Lubetz - Arthur Lubetz is an Adjunct Professor in the
School of Architecture. He brings a theoretical mindset, a creative
framework, and a rigorous approach. He is also the fall semester
instructor. I have not collaborated with Arthur before though
he once taught a parallel studio. One of Arthur’s key driving
principles is the inclusion of the body in architecture. This relates
closely to my thesis.
aD V I s o R s a n D PR I m a R y Co n T a C T s
Fig. 1.3 My Fall 2010 studio project that Art Lubetz critiqued and reviewed.
1 0
Dale Clifford - Dale Clifford is an is an Assistant Professor in
the School of Architecture. He has significant background finding
simple solutions to complex problems using media not native to
the problem. I have had Dale in two previous classes, Materials
and Assembly and BioLogic, both of which involved combining
disparate systems of assembly to achieve a goal not easily or
impossibly reached by any constituent system. Dale may also
provide many connections into digital fabrication practices.
Joshua Bard - Joshua Bard is an Assistant Professor in the
School of Architecture. He should contribute some digital and
media expertise. He was be the spring semester instructor.
Joshua co-taught a fall course, Parametric Modeling (the
other instructor being Ramesh Krishnamurti) that focuses on
integrating a software with Rhinoceros, Grasshopper, although
that software is built inside Rhinoceros as a plugin. Joshua may
help with adapting other software.
Ramesh Krishnamurti - Ramesh Krishnamurti is a Professor in
the School of Architecture. He should contextualize my thesis due
to his background studying computer visualization and vision. He
taught a course I took in the spring, Parametric Modeling. I have
worked as a Teaching Assistant with him for the class Descriptive
Geometry for a few years. He is also a great thinker - he may help
me work out the nature of my thesis and any kinks it might have.
Fig. 1.4 Samples of work made in Materials and Assembly (MnA), BioLogic, and Parametric Modeling. Top to bottom: The MnA enclosure made with zip ties; A
responsive wall using nitinol; Parametrically defined surfaces.
1 1
Varvara Toulkeridou - Varvara Toulkeridou is a graduate
student in the School of Architecture. I have worked with her
while being a Teaching Assistant for Descriptive Geometry under
Ramesh. As she has a similar background and knowledge to
Ramesh, she may be another useful source of advice and critique.
She is also currently a Teaching Assistant in the Parametric
Modeling course that I am taking, making her available weekly
should I have specific questions I need to ask her.
Kai Gutschow - Kai Gutschow is an Associate Professor
in the School of Architecture and was the fall and spring thesis
coordinator. He developed the program as it ran, and managed
all of the students’ time and projects. He coordinated with
Johnathan Kline and Mary-Lou Arscott, the Associate Head, to
prepare the final presentation location at the Miller Gallery.
Johnathan Kline - Johnathan Kline helped Kai during
the spring. He established a more lax schedule for the spring
semester compared to the fall semester. He also organized some
of the early meetings and reviews the thesis students participated
in early in the spring semester.
1 2
PR o j e C T bR I e fThe architectural render has long been the pinnacle of drawn
design - a constructed image that shows the viewer an idealized
view of an architectural project from a specific location within
the project at a specific time of day. Traditionally, the architect’s
primary tool for image-making was a drafting board. Some time
in the last few decades architects have adopted the computer
to serve the same role yet advance it in many ways, making the
digital render an evolution over what was possible with drafting.
Yet, despite the apparent approach towards a visual quality near
that of human sight, the digital render failed to fully use the full
power of a computer. The digital render took a horse cart and
made it into an automobile but failed to then also make a van, a
truck, or even a race car.
The allure of a digital world has fascinated people ever
since computers were able to create early vector and later raster
graphics. The idea has been explored in such films as Tron (1982)
and The Matrix (1999) and more recently in Avatar (2009), where
over half of the film was photorealistic computer effects, as
well as hundreds of student or collegiate art projects. It has led
to the development of hardware to augment the human frame,
extending what the human mind is limited to by the body. Digitally
Fig. 1.5 The complete toolset in Rhinoceros for animations.
Fig. 1.6 Diagram created by a developer of Brigade 3, a cutting edge path renderer, made by OTOY, the same people behind Octane. It posits that, after a
certain amount of geometric detail, ray tracing always beats raster meshes.
1 3
Fig. 1.7. Three approaches to my thesis. Top to bottom: taking a render, creating many renders from it, then showing them together as an animated sequence under the control of the viewer, faster than just a series of renders; The render and the model are combined into a visual system whereby the user can explore the model in a virtual world, allowing her or him to share the model with anyone; With a real time render the concept of presence comes into play, since a moving realistic image
allows the viewer to inhabit the image.
fabricated films have gradually replaced hand-drawn films and
have even entered the mainstream as a respected category of
film. Architectural designers have tapped this field, but not as
fully as they could have.
Another way the digital world has entered the social
consciousness is through video games. While not all video games
involve a 3D virtual environment, the ones that do often go for a
highly photorealistic portrayal of a digital environment. The tools
video game designers use are often made specifically to quickly
develop virtual environments. Students have often tried to use
such tools in their projects, but although they tended to gain
success architectural firms have rarely followed suit.
It is true that video game designers create objects that are
meant for mass production, and film companies make objects
meant for mass exposure. This kind of thinking dodges the aim
of my thesis though, because I am not proposing architecture
become video game-like or film-like. I am proposing it use the
tools they use maximize digital communication.
+ + + + + + + + + +
The thesis is a field produced by two axes - the vertical
axis is that of architectural image-making: how have designers
evolved their tools to match current technological advances; the
horizontal axis is that of digital interfaces and interaction: more
1 4
and more society is finding ways to interconnect with itself - such
interaction in architecture, a field entirely involved in the business
of being around others, seems largely absent or unused.
The first axis, visualization:
While many designers in the field have advanced the static
render into something more dynamic, making videos or flythroughs
or virtual habitats, more often than not these cases were one-
time gimmicks and have not established as a versatile aspect of
architectural design.
The second axis, interaction:
The concept of digital interaction has often been explored by
artists trying to cope with the digital frontier yet the possibility of
delivering an architectural project with extra-sensory exposure does
not seem to have gained traction among architectural designers,
even though technology exists to allow interaction beyond that
which is seen or heard.
The project, therefore, is to explore and define the extent
of such efforts in both directions, identify what was tried, what
failed, and how those attempts could be improved, identify the
best candidates (by an evolving criteria as the project develops)
for a concentrated push into versatility, and produce a working
example of the next evolution of drafting.
Fig. 1.8 Is there a possibility here?
Fig. 1.9 THE GRID. Neither interaction nor visualization alone will achieve any greatness; it is through the collaboration of the two axes that a far greater
advancement can evolve.
1 5
The primary deliverable will be a software package which
parallels or replaces the point in design when a designer of
architecture would make a static render and, instead of producing
a mere digital render, would create an interactive simulation
serving as proof of experience much like an architectural model
is a proof of assembly.
A distinction has to be made between a pre-rendered
animation and a real time interactive environment. While pre-
rendered animation is a side-effect of this under-utilized function
of computers, it is absolutely a rut of possibility. It is a linear
evolution of a digital render - why stop there when a render can
evolve planarly?
+ + + + + + + + + +
A breakdown of the thesis into one sentence, three short
sentences, and a short paragraph is a useful tool for understanding
the thesis:
1: To Seek a Means and the Benefits of a System to Interact in
Rendered Real Time With Digital Models.
3: Such a system would provide architects and clients a
preview of the visual and aural aspects of a building in their
entirety before the building is built. Much like how a physical
model is a proof of assembly this would be a proof of experience.
So what?
Fig. 1.10 An example of a virtual environment that can be explored. It is both dynamic and interactive - it goes beyond what a set of renders could have done and also gives the user something a render could never have - a sense of presence in
the project.
1 6
9: Architects traditionally make analog products - visual
stimuli that mimic the rays of light that true sight gives. For
presentations (renders) and data analysis (orthographics), these
products are nearly always static images. Yet, much of architectural
design requires the input of a user’s movement to activate. No
static image will ever describe to the designer the experience
of natural movement within a project. Without an interactive
experience to iterate from, the final, built, experience cannot
be prototyped. Interpreting a static image requires a skill called
mental rotation that is learned through studies of descriptive
geometry, long exposure to architectural orthographics, and
CAD. Mental rotation is a skill not every client has and not every
architect develops fully. Without this skill static images become
severely lacking because too much of the design process relies on
interpreting these images with the aim of improving the design.
Opportunities exist to replace or compliment static images with
real time renders that closely resemble the built design both
experientially and conceptually, which would allow a more in-
depth design pipeline.
1 7
Research - the first step of the thesis was to generate
a foundation of knowledge in the field of visualization and
architectural visualization in particular. The thesis combines
several schools of thought - Representation, Automation
through Technology, Simulation, Video Gaming, Interfaces, and,
naturally, Architecture. Each field contains several informative
areas: History, Technology, Application or Practice. These areas
informed what was available in the field as well as dictated
possible constraints. For a broad spectrum I expected at least
six established literary sources and six other collateral sources
(videos, talks, examples of work).
Definition - in the meantime, I continued to refine the
grounds of my thesis - the product, the deliverable, is a tool. The
means is often more important than the end because the means
is inherently repeatable. The research molded the form and
function of the thesis and its ultimate deliverable, a visualization
tool.
PR o j e C T me T H o D s a n D T I m e l I n e
DDDDDDDDDD
RRRRRRRRRRRR
EEEEEEEEE
Sept
emb
erO
ctO
ber
Sep. 3 - Version 2 of Thesis
Sep. 9 - Version 3 of Thesis, focus on methods
Sep. 16 - Version 4 of Thesis, expand on all sections
Sep. 18 - Version 5 of Thesis, presented as a poster
Oct. 21 - Version 6, review
Oct. 4 - List of deliverables
Oct. 18 - Midsemester break
1 8
Experimentation and Evaluation - the second step was
an exhaustive analysis of existing visualization software (or
hardware, if it is available through CMU) for the purpose of design
(NOT making a final product but as another step, or a better step,
in an iterative process). This involved its own research on what
tools architecture firms have used in the past (and documented)
for time-based deliverables and subjectively evaluate them based
on those deliverables. Following research on what tools practicing
architects used, I performed research on tools students have
used, what artists of various caliber have used, and video game
engines. While the time each visualization tool takes to render
(from hours per frame to frames per second) is crucial, I also
looked for other design features, keeping the root of my thesis
in mind - the possibility for the digital real time. Theoretically this
research could have come across examples of work, but the focus
was on how those were made, not what they were.
Oc
tOb
er
DDDDDDDDDD
RRRRRRRRRRRR
EEEEEEEEE
NO
vem
ber
Dec
emb
er
Nov. 28 - Thanksgiving
Dec. 8 - Review of thesis development
Dec. 13 - Submittal of thesis book
Dec. 16 - Last day of first semester
1 9
Compilation - The two threads of research combined. At this
point I planned to have a steel-hard definition of my thesis. There
would have been at least two deliverables, one for each body of
research. The literary deliverable would have been an opinion
piece drawing from all the sources I compiled that projected the
possibility (that I believe is the case) of what architects could
embrace in the field of visualization given the power of computers
and what effect it would have on current design paradigms. This
opinion piece would have predict the possibility of the second
deliverable. The software deliverable will be a proof of concept or
a redistributable software package (depending on if the software
I end up choosing is licensed for educational use or distribution).
This software package would have supported the opinion in the
first deliverable, ultimately proving architects can evolve the
render into something that interacts on a level above the visual
or tactile.
The software package would have addressed the range of
interactivity that is missing in architectural delivery. Depending
on what software I used, there would have always been a way for
both the client and the designer to enrich their communication.
The software package was, necessarily, an all digital item, as
having a video or a screenshot of it defeats the point of interaction.
EEE
DDDD
CCCCCCCCCCCC
JaN
ua
ry
Feb
ru
ar
ym
ar
ch
Jan. 13 - First day of second semester
Jan. 20 - MLK Day, no classes
Mar. 7 - Spring Break starts
Mar. 5 - Midsemester thesis review
2 0
CCCCCCCCCCCC
Beyond - if there were yet more time I may have develop
more deliverables to parallel the two main deliverables in the
Compilation step. One would have been a documentation on the
use of the software package and tool. A certain amount of basic
tutorials smoothening the learning curve would already have
been part of the software deliverable, but, like any software,
much of the tool would have been difficult to approach for a
new user. If there were time I could have developed detailed
explanations of various functions within the software package.
Importantly, this would have heavily depended on the nature of
the software package. If it were a video game engine editor then
it may have grown to have dozens of tutorials. If it were a small
utility (perhaps an architectural firm has developed one), then
there may only have been a small handful.
Mar. 17 - Spring Break ends
BBB
apr
ilm
ay
ma
rc
h
Apr. 10 - No classes for Carnival
Apr. 13 - Carnival ends
May 2 - Last day of classes
Apr. 25 - Final Presentation
May 12 - Thesis due
2 1
Precedents are difficult to find because the bulk of
professional architectural animation focuses on pre-rendered
scenes. The videos that are produced by companies that focus
on this kind of animation are often flythoughs or disembodied
gliding camera views moving through completed designs, either
as part of a submission to a competition or after the design was
built.
The short Wikipedia page on architectural animation
mentions how difficult it is to render animations and how firms
rarely have access to the hardware or tools to assemble such
products. However it also mentions that, more and more, firms
are recognizing that animations are better at conveying the ideas
of a project than design diagrams. Otherwise, there seems to be
little effort anywhere to document the most effective animations
or even any attempts at real time interaction with animation.
Two companies exist that have begun using game-like
software to create virtual versions of architectural projects.
Both focus on Unity3D and create services ranging from training
simulations to marketing packages. Both companies have
harnessed Unity3D’s ability to work cross-platform as well as its
ability to efficiently handle a complex scene with pre-computed
PR e C e D e n T s
R e s e a R C H
Fig. 2.1 Arch Virtual’s web version of one of their projects.
2 2
Fig. 2.4 A deliverable from Real Visual on a mobile platform.
Fig. 2.5 Arch Virtual’s Unity3D booklet and their application of the Oculus RIft virtual reality headset.
shadows and materials.
The first company, Real Visual, focuses on a high quality
of delivery in simulations, training, marketing, and outsourced
design work. They cover work in various multi-national sectors
aside from architecture - energy, transport, and defense. This
displays flexibility and expandability, and shows how such
technology and its application are quickly burgeoning in the
wider world. They work closely with the developers of Unity3D
to ensure the software is as cutting edge as possible. If architects
could learn from the technical expertise of this company then the
field would only be enriched.
The second company, Arch Virtual, focuses more on cutting
edge hardware and integrating it with Unity3D. They have
worked with the Oculus Rift, a virtual reality headset currently
in development, bringing in projects developed in Unity3D, that
are also configured to work on mobile platforms like those of
Real Visual, and setting them up to work with the headset. They
also have an ebooklet detailing the steps required to create an
architectural project within Unity3D. This booklet is a step in the
right direction for the profession, but it is by far not enough, as
at 65 pages it is only a set of guidelines rather than thorough
educational materials.
Autodesk also has software designed for the purpose of
accelerating architectural animation. I mention this not because
Fig. 2.2 Real Visual’s logo.
Fig. 2.3 Arch Virtual’s logo.
2 3
it is an effective precedent for my thesis but because it is exactly
the wrong approach - it does not use a human viewpoint, it
does not offer a high level of realism in its graphics, and it favors
presentation over interaction.
This software, Autodesk Showcase, takes models and allows
the user to dress them up, applying materials and environments
to the scene. It offers various alternate rendering types, like
cartoon or sketched, as well as options for sets of materials to be
shown by themselves. The workflow is one of setting up renders
or animations with a preview viewport and then rendering them,
akin to what a full screen Vray would look like.
The biggest drawback that I perceive in this software is,
despite its effort to offer architects a more intuitive rendering
solution, that it fails to advance the field. It is an example of
stagnation: nothing in it is radically new over what is already
possible in AutoCAD, Maya, 3DSMax, or Rhinoceros with Vray. It
is a horizontal advancement and fails to use advanced rendering
methods, new interaction methods, or take advantage of newer
hardware.
+ + + + + + + + + +
It is also important to note video game graphics precedents.
There is a stigma within the commercial culture today that
video games and their technology are beneath professionals
Fig. 2.6 Autodesk Showcase screenshots. Clockwise from top left: Regular preview view; Cartoon preview; Different material sets; Publishing, or rendering,
an image.
Fig. 2.7 Autodesk Showcase logo.
2 4
Fig. 2.8 QR code for a demo video of
CryEngine.
Fig. 2.9 QR code for a demo video of the
Fox Engine.
and their interests. While it is true that the gameplay aspects
of video games have little bearing in most professional fields,
the technology and simulation aspects behind video games are,
by now, entirely applicable in other fields. (There are also such
things as GWAPs, games with a purpose, video games designed
specifically to be training materials and high-fidelity simulations
of real-world scenarios).
For the purposes of my thesis I will argue that the graphics
advancements of video games have, over the past several years,
reached such high levels of realism, among the video games that
use cutting edge engines, that they contend with professional
rendering software in terms of speed, quality, and production
value.
Modern video game engines generate lighting and shadows
dynamically, meaning there is no pre-computation except that
which is necessary to place the geometry into the scene. For
materiality many games still use shaders, simplifying computation
and sacrificing some real time effects, but some engines have
begun developing real time shader effects, namely refraction and
reflection.
As for simulation, all video games with a first person
perspective already have immersive interaction and exploration,
key features for visualization that are lacking in professional
software packages.
Fig. 2.10 Super Mario 64, not an example of a contemporary video game.
Fig. 2.11 Crysis 2, an example of a contemporary video game using realistic graphics.
2 5
Fig. 2.12 Mind Map as of October 21st, 2013. My thesis subject area is in the top left corner.
Fig. 2.13 QR code for the mind map.
The literature took up the bulk of the work during the first
half of the first semester after the thesis program got started.
The literature review pulled from over 30 sources more than a
third that provided valuable insight into the context of my thesis.
This proved to peers that this is an academic subject and bears
worth in the field of architecture. The very shadowy nature of the
subject of my thesis is exactly why I am proposing my thesis - to
raise awareness of what can be done with modern tools.
I also created a mind map, open to my advisors to flesh out,
as I continue to insert data siblings and children. The mind map
charts everything in the field of computing that could relate to
my thesis - it is an attempt to contextualize my work, to bring it
from computer science to a position that is understandable by
architects. The semantics of my thesis automatically raise various
stigmas in readers or reviewers, so having a way to visually place
my thesis among other academic subjects is important.
Ideally, any interaction with computers that architects could
have should have a spot on this mind map and right now my thesis
only occupies a small portion of it. But one of the points of my
thesis is that this should not be so. Interactive visualization can be
a powerful ally in developing a design, and by expanding that field
l I T e R a R y Re s e a R C H
2 6
architects could learn more powerful, more flexible tools.
Following is literature research, a review of several sources
that bring up important points for visuality and rendering as they
apply to architecture:
Visual Digital Culture: Surface Play
and Spectacle in New Media Genres
This book by Andrew Darley explored visuality and spectacle
in digital media. I drew parallels in it with architecture with how
early digital modeling was show-driven - digital rendering is
often about what a project could be like and not what it is. The
greater definition of model, to simulate, comes into context,
showing how lacking static renders are. It also showed how video
game software could be photorealistic, an important point that
I must continually clarify. The illusion and wanting to be fooled,
repetition and customization, the sense of occupancy and a
comparison between video games and virtual environments
rounds out the content of the book.
Relevant quotes in textual order:
“A key example of such research was that into real-time interactive computer graphics. This came to practical fruition in 1963 in a system called Sketchpad, which allowed a user to draw directly on to a cathode display screen with a ‘light-pen’ and then to modify or ‘tidy-up’ the geometrical image possibilities so obtained with a keyboard. Though extremely primitive by today’s standards, Sketchpad is viewed as a crucial breakthrough from which have sprung most of the later technical developments in the areas of so-called ‘paint’ and interactive graphics systems. By the mid -1960s, a similar system involving computer image modification was being
used in the design of car bodies - a precursor of current CAD/CAM (Computer Aided Design/Computer Aided Manufacture) systems. And by 1963, computer generated wire-frame animation films -visual simulations of scientific and technical ideas -were being produced using the early vector display technique.” - pg. 12
This is significant as a historical precedent on the type of
interactive software that my thesis belongs to. Sketchpad, Ivan
Sutherland’s own thesis, was the grandfather of CAD modeling.
While it crucially combined hardware and software, within the
realm of modern software and interface systems my thesis does
not have to have the same intertwined nature. Ideally my thesis
should be able to do everything with a keyboard and mouse,
however exploration into alternative hardware input is possible.
The point is to separate the tangential, relatively, development of
software like REVIT and AutoCAD from this original thread.
“The desire on the part of scientists to model or simulate physical processes and events in space (and time) was a central impulse in the production of the earliest computer graphics and films. Whilst concurrent with the initiation of applied forms, work was under way on computer produced figurative imagery as a research activity in its own right. Even the work conducted in collaboration with artists had a decided leaning towards more figurative kinds of imagery. At the end of the 1960s experimentation began into the production of algorithms for the production and manipulation of still, line-based figurative images.” - pg. 14
The notion that early computer graphics were, in a way,
show-driven, relates well to how architects do things with
technology. Architects often use computers and rendering to
show what a project could be like, as opposed to showing what
it actually is. The original scientific drive to model, however,
encompasses more than just showing the project itself, but also
showing what the project could do. Here the greater definition
Fig. 2.14 Visual Digital Culture: Surface Play and Spectacle in New Media Genres cover.
2 7
of ‘model’ applies, in that ‘to model’ means ‘to simulate’, where
various possibilities enter the game and a static representation
becomes lacking.
“The one that came to discursive prominence within computer image research and practice is perhaps the one with which we are all most familiar. Quite simply it turns upon the notion of the proximate or accurate image: the ‘realisticness’ or resemblance of an image to the phenomenal everyday world that we perceive and experience (partially) through sight. For the majority of those involved with digital imaging at the time, the yardstick of such verisimilitude was photographic and cinematographic imagery.” - pg. 17
This is another thing to keep in mind, while my thesis may
include video game software an important benchmark is that I
do not sacrifice photorealism. I am mentioning this because one
aspect of my thesis is that it takes several steps forward, and very
few, if any, back.
“In this case, of course, the set is virtual or latent - itself a simulation created and existing in the program of a computer. Such programs are now able to simulate three dimensional spatial and temporal conditions, natural and artificial lighting conditions and effects, surface textures, the full spectrum of colours, solidity and weight, the movement of objects and, as well, the complete range of movements of a camera within and around their virtual space. When cartoon characters - and, just as important, cartoon tropes such as anthropomorphism - are imaged through this studio simulacrum, then new registers of mimetic imagery are achieved within the cartoon: a consequence of this peculiar crossing or fusing of traditionally distinct forms of film.” - pg. 85
A parallel discipline to my thesis is digital film animation.
With digital film animation, the software technology is, by
necessity, highly configurable and allows total control of a
virtual scene. While such control is not applicable to architectural
design because the digital in architecture is merely a step in the
development, seeing what is possible in the field will allow me to
find an upper bound in software capabilities.
“A technical problem - the concrete possibility of achieving ‘photography’ by digital means - begins to take over, and to determine the aesthetics of certain modes of contemporary visual culture. Attempts - such as those focused upon here - to imitate and simulate, are at the farthest remove from traditional notions of representation. They displace and demote questions of reference and meaning (or signification) substituting instead a preoccupation with means and the image (the signifier itself) as a site or object of fascination: a kind of collapsing of aesthetic concerns into the search for a solution to a technical problem.” - pg. 88
This is the other side of the problem. Attempting to focus
too much on the signified versus the signifier may break the
relation of the image to the model or what it is modeling. The
effort to produce a visually realistic image moves too far from the
ideal that the task of creating the image in the first place started
off from - in visual representation that ideal is to show truthfully
what the virtual environment looks like, and in architecture and
my thesis that ideal is to show a model experientially - through
space and time.
“This involves surface or descriptive accuracy: naturalism. At the same time as distinguishing itself as other (alien) in relation to the human characters and the fictional world, the pseudopod must appear as indistinguishable at the level of representation, that is to say in its representational effect. It had to appear to occupy - to be ontologically coextensive with - the same profilmic space as the human actors. This involved the seamless combining of two differently realised sets of realistic imagery: of which one is properly analogical, i.e. photographic, the other seemingly photographic, i.e. digital simulation. Additionally however, it must also integrate, again in a perfectly seamless manner, into the diegetic dimension: the story space. In order for this to occur an exceptional amount of pre-planning had to enter into the carefully orchestrated decoupage that eventually stitches the shots together. Here, finally, surface accuracy is subordinated to the rather different codes of narrative illusionism.” - pg. 108
Here the author was analyzing a scene from the film The
2 8
Abyss (1989) where a computer generated tentacle is made to
coexist within the filmic space with the real characters and setting
and also within the presentational space, where the story as shot
has to make room for this element which will be added later in
the production of the film. The importance of this is again that
the purpose of a render, or real time interaction, is not the pretty
image itself but what the image does, its performative element.
The quality and the believability of the frame in a film example
has to kneel to the frame as a narrative element - this tentacle
in The Abyss has to make sense as a tentacle first, the image of a
tentacle later. Likewise in architectural representation, an image
of a project has to come after what the image will do, which is a
proof of experience.
“The contradiction - ever present in special effects - between knowing that one is being tricked and still submitting to the illusory effect is operative here. Yet, particularly (though certainly not solely) in those scenes involving computer imaging discussed here, the more photographically perfect or convincing the images, the more - paradoxically - does their sutured and suturing aspect seem to recede and their fabricated character come to the fore.” - pg. 113
This pertains to the effect of illusion and wanting to be
fooled. Sometimes a fabricated image, a computer generated
mosaic, becomes too artificial. This is important to note because
it is possible that so much effort can be spent on making an
architectural image perfect photographically that its photorealism
eclipses its narrative - its experiential conduit. Just like there are
technological functionality bounds - software exists that can do
many, perhaps too many, things in a virtual environment - there
are aesthetic bounds - software cannot be so focused on being
realistic that the realism gets in the way of the representation.
“It is both the bizarre and impossible nature of that which is represented and its thoroughly analogical character (simulation of the photographic), that fascinates, produces in the viewer a ‘double-take’ and makes him or her want to see it again, both to wonder at its portrayal and to wonder about ‘just how it was done’.” - pg. 115
This, on the other hand, produces a lower bound on the
aesthetics of the image. It is likewise cautionary to make an image
too experiential, too generative of wonder. The combination of
seemingly impossible imagery rendered (by computer) with
accurate realism, so to say, produces a kind of inquisitiveness
that places the generation of the image itself before what the
image represents. The way the image was made becomes more
interesting than what the image is about.
“Thus the fact that we can make many identical copies (prints) of a particular film, means not only that more people get to see it but also that as a work it is thereby made less precious.” - pg. 125
This passage refers to Walter Benjamin’s theories on
mechanical reproduction. It is always a good idea to keep in
mind the fact that quantity, even if it maintains quality, does
not necessarily increase the popularity of a work. Since a part of
my thesis is to explore if architectural simulations can become
portable, it will be important to see what effects such mobile
qualities have on architectural design.
“today it is not what is repeated between given tokens of a series that counts for spectators, so much as the increasingly minimal differences in the way this is achieved. Burgeoning ‘replication’, the repetition at the heart of commodity culture, forestalls the threat of saturation and exhaustion by nurturing a homeopathic-
2 9
like principle of formal variation (i.e. based on infinitesimal modifications and changes).” - pg. 127
The issue of repetition versus customization further explores
what architectural representation could become in a mass mobile
environment. This particular passage refers to the phenomenon
of television shows, comic strips, and serial novels where only
small changes are made between versions, only enough so that
a new installment is different from the last. Theoretically the
proliferation of architectural representation into the mainstream
could go this way - an architectural firm produces an interactive
architectural simulation or a few, and a client modifies it only
slightly. Perhaps that is an unideal future.
“Even fields such as computer games and simulation rides, which are the most recent and appear to depend more on the novelty of the technology itself, are - as we shall see in coming pages- just as much subject to this aesthetic of repetition. They may involve new formal elements - the much vaunted ‘interactivity’ and ‘immersion’, for example - and these may well affect their individual aesthetics. However, just as much as the more established forms, they also seem destined to operate within the logic of self-referentiality and the preponderance of the ‘depthless image’. All are manifestations of an altogether new dimension of formal concerns that established itself within the mass cultural domain of the late twentieth century, helping to constitute both cultural forms and practices of production and aesthetic sensibilities.” - pg. 129
Here the author combined the two threads of thought -
repetition of the image in culture and a focus on the image itself
over the substance of the image. The idea here is that as an image
spreads it does not necessarily mean that people see it more, or
see through it more. The proliferation of an image may shift the
audience’s concern towards the formal quality of the image, put
another way, more people see less. Being able to have a large
audience for an image may be a large factor - in an architectural
firm and with a client only a small number of people see the image
and can control it - once such limitations are lifted, if they can be
lifted, the image may be diluted even if it gains other properties,
like interactivity.
“Living in cultures in which we are surrounded on all sides by moving images, we are now particularly accustomed to the kind of montage that strives to hide its artifice.” - pg. 131
Architecture is, independent of what some architects think,
part of the global digital stage and as such has to compete with
other visual fields. The more graphically advanced the rest of our
culture becomes, the more certain qualities will be expected of
the visual elements of architecture. This means that fleshing out
this aspect of architecture, or at least exploring it in my thesis, will
also require me to know what is expected of real time interaction
as well as what it can do.
“The sheer sense of presence, however, conveyed in the best of them - and here Quake is a key example - compensates for such defeats. In other words, it is the experience of vicarious kinaesthesia itself that counts here: the impression of controlling events that are taking place in the present.” - pg. 157
Here the author brings in the experience of video games,
saying how, in the interaction with the game, the fact that the
player may sometimes need to repeat areas in a video game is
overshadowed by the fundamental fact that the player is actually
controlling something in the virtual realm. This is an aspect of real
time interactive simulations that needs to be put in the forefront
because it simply does not exist in renders or even CAD programs.
3 0
There is no sense of time in Revit or Sketchup, and watching an
animation gives the user no control. While substance is key in the
image, presence is important outside it.
“interactive representation involves a mode of representing that is ‘inside the time of the situation being described’. That is to say, time is represented as viewed from a first person perspective - literally as if one were really there, thereby, producing the impression that things are continually open to any possibility...Indeed, it becomes difficult to untangle space from time in this respect so intimate is their relation. We might say that the illusion of experiencing events as if they are taking place in present time in computer games is largely dependent upon visual simulation.” - pg. 158
Here the author points out that the mere introduction of
time to a virtual environment already creates the impression of
interaction by the simple virtue of providing limitless possibilities
on ‘what could happen next.’ In video games, the visual alone
can do this. Likewise in my thesis, establishing this effect by
the photorealistic representation of architectural models could
already be a huge step towards interaction.
“given the increasing surface realism of the moving imagery, the sophistication of real-time graphic representation and the use of first-person perspective, the impression of actual occupancy and agency within the space of the game’s fictional world can be extremely convincing.” - pg. 163
Another aspect of video games that can be transferred to
interactive architectural simulation is the sense of occupancy.
Through a combination of realistic imagery, realistic depth
(material effects and believability of presence), and a simulation
of what it would be like as if one was there, occupancy can be
achieved. Since occupancy is a major aspect of experience, such
a conceptual framework is important for the field of my thesis.
“However, such ‘active participation’ should not be confused with increased semantic engagement. On the contrary, the kinds of mental processes that games solicit are largely instrumental and/or reactive in character. As I suggest above, the space for reading or meaning-making in the traditional sense is radically reduced in computer games and simulation rides.” - pg. 164
Here the author steps back and concedes that the actual
interaction with a video game is not the same thing as interaction
with the virtual environment. The user is still fundamentally
looking at an image. This is also very important to keep in mind
because my thesis does not seek to redefine how architecture
is made - it seeks to augment or improve only the computer
representation aspect of architecture.
Generating Three-dimensional Building Models
From Two-dimensional Architectural Plans
The only relevant quote:
“The building model used to develop and demonstrate the system was produced by iteratively applying “clean-up” algorithms and user interaction to convert a grossly inadequate 3D AutoCAD wire-frame model of Soda Hall (then in the design stages) into a complete polyhedral model with correct face intersections and orientations. The Berkeley UniGrafix format was used to describe the geometry of the building, because of its compatibility with the modeling and rendering tools available within the group. The interior of the building, including furniture and light fixtures, was modeled by hand, through instancing of 3D models of those objects. In all, the creation of the detailed Soda Hall model required two person-years of effort. It became clear that better modeling systems were needed.” - pg. 3
While the research report, by Rick Lewis, was written in
1996, before significant advances in CAD had taken root among
the designing audience, the general gist of what this quote refers
Fig. 2.15 Generating Three-dimensional Building Models From Two-dimensional Architectural Plans
cover.
3 1
to remains true today. With my thesis this argument would more
pertain to having to customize every render for a flawless end
result (presumably), the notion that accurately modeling an
entire building in a computer is manually labor intensive is true
- partly because many designs are so unique, there are no tools
for efficiently spreading geometric complexity within a model
without resorting to grids or simple patterns. With rendering and
interaction, the manual difficulty lies in preparing a render scene
and then setting lighting and material properties, all of which take
a large percentage of the total time it takes to develop a render.
Perhaps there is a way to develop a pipeline where materials and
lighting can be more easily established without thinking of it as a
necessary preparation for each render scene.
Visuality for Architects: Architectural Creativity and
Modern Theories of Perception and Imagination
This book by Branko Mitrović introduced an idea to my thesis:
mental rotation, the ability to rotate a 2D representation in the
mind. It bashed architects for blindly relying on narrative as the
prime way of communicating projects and designs. It proposes
that architecture evolve into a visual profession. Generally, it
noted a behavior in architects to avoid or ignore architecture’s
purely visual aspects. The idea of ideological bias versus the
opportunity to see architecture visually is critical to expanding
the use of interactive media in architecture, yet architects first
need to open their mind to the notion that architecture is not
narrative by default.
Relevant quotes in textual order:
“What psychologists describe as mental rotation is the same kind of task that is performed by computers in modern architectural practice.” - pg. 6
This book argued that what CAD does is not fundamentally
different from what a human brain does when it views a plan
or a perspectival image - though the separation of conceptual
thinking from visual thinking becomes easier in a computer.
Thus relying on creating static images just so the brain can be
forced to have visual and conceptual thinking near each other,
forcing connections, is a fairly outdated concept - the process
can be separated, CAD can give the full visual stimulus that real
experience provides with a real building and the brain can be fully
used for conceptual thinking.
“The same tendency to base design on stories that can be told about architectural works is common in contemporary architectural practice as well. Here it is strengthened by the fact that in order to get commissions, architects often have to explain in words their design decisions to their clients. Sometimes they (are expected to) invent stories about what the building represents.” - pg. 11
Another key theme the book brought up was the stubborn
reliance of contemporary architects on narrative and having, or
thinking that it is the only way to, describe a building’s ‘concept.’
Why rely on speaking about an almost inherently visual idea
(granted, tactility and sound matter) when you can communicate
it visually?
Fig. 2.16 Visuality for Architects: Architectural Creativity and Modern Theories of Perception and
Imagination cover.
3 2
“In fact, much bigger issues are at stake. Architecture does not live in isolation from its intellectual and cultural environment. If antivisual biases are going to be credible among architects, architectural academics, or theorists, this can happen only if such views are based on and derive from assumptions that are credible in the society in which they live.” - pg. 13
Socially, one can argue that the visual has grown faster
and faster in developed society. Take the internet - experienced
almost exclusively visually: computer screens, smartphones,
tablets, even printouts of web content are visual objects. Film,
video games, advertising, it is all visual. Perhaps even literature
financially is falling behind visual storytelling through film,
TV, Netflix, and so on. Therefore architecture must develop,
somehow paradoxically, into a visual profession. That is nearly at
the core of my thesis.
“Applied to architecture, this means that there are no visual properties of architectural works that are not ultimately derived from the ideas we associate with these works. Visual perception of buildings is merely a result of the knowledge and beliefs we already have about them.” - pg. 14
A bit of theory here. The more the brain is forced to draw
from its reservoir of constructible memories, when exposed to
a single image of a piece of architecture, the more the brain will
generalize to the archetype. The brain, when it has to make up
information, will just use what it already knows. Thus it is in fact
detrimental to the review or design of architecture if people view
it in a reduced manner, that is, in a manner far from the actual
experience of architecture. I propose that a greater reliance
on interactive visualizations, being that those are closer to said
experience, would promote a truer review of architecture.
“If we are going to talk about the aesthetic qualities of architectural works, we need to be aware that these works are going to be thought about not only as perceived from a single point in space but as three-dimensional objects. We perceive a building from one side, from another, from inside, we observe the composition of spaces, and after some time we have formed a comprehensive understanding of the building’s three-dimensionality. Or, we don’t have to be dealing with a built building at all; we can grasp its spatial qualities by studying its plans, sections, and elevations. By analogy with 3-D computer modeling, one could say that we have formulated a 3-D mental model of the building in our minds” - pg. 71-72
Again with mental rotation. Much of architectural experience
revolves around understanding the visual composition and
relationships of a design or building. This is possible from a human
vantage point with a built building, but with design products, the
observer has to effectively rebuild the model inside their mind.
It would only accelerate the understanding if the observer could
interpret something only a step away from actual experience, an
interactive render.
“In a situation where it is recognized that architectural works can be perceived, imagined, thought about, mentally rotated, and that their geometries can be studied, their colors discussed, and so on, independently of any concepts or meanings we associate with these works, only an ideologically biased professor can insist on evaluating the work exclusively on the basis of the story that can be told about it.” - pg. 85
This pertains to the general issue where architects are not
grasping the full breadth of the tools that are available to them.
The somewhat hesitant reliance of architectural reviews to
generalize renders to drawings paired with a reliance on printed
material is stifling architectural design flexibility. Thus, in an effort
to justify their views (ironic), review boards pretend that they are
in fact not interested in the visual and are looking for (inescapable
3 3
irony) a more narrative description of the project. The idea
of ideological bias versus the opportunity to see architecture
visually is critical to expanding the use of interactive media in
architecture.
One Approach for Creation of Images and Video
for a Multiview Autostereoscopic 3D Display
This research report by Emiliyan Petkov outlines a method
for creating images for 3D screens, useful to know for my thesis.
A relevant quote:
“A matter of interest is exploring the possibility for developing interactive applications for 3D displays. This kind of applications gives users the opportunity to interact with objects in a computer simulated world in real time. Thus the time for remaining in this virtual environment is not limited and decisions what to do and where to go are made by the user. These applications will offer an opportunity for creation of virtual worlds through the multiview autostereoscopic 3D displays.” - pg. 322
Somewhat tangential, part of my thesis is exploring possible
hardware for interaction, one of which would be 3D Displays
or Monitors or Screens. A strong aspect of that would be, not
just review of the design using this hardware, but also creation,
potentially collaborative.
Touchable 3D Video System
This research report by Jongeun Cha, Mohamad Eid, and
Abdulmotaleb El Saddik introduces the idea of presence - the
immersive feeling of being inside a virtual environment.
Relevant quotes in textual order:
“Recent advances in multimedia contents generation and distribution have led to the creation and widespread deployment of more realistic and immersive display technologies. A central theme of these advances is the eagerness of consumers to experience engrossing contents capable of blurring the boundaries between the synthetic contents and reality; they actively seek an engaging feeling of ‘being there,’ usually referred to as presence.” - pg. 29:2
In the entertainment industry, displays are getting larger and
larger, with more accurate color rendition and higher contrast
ratios - this is driven by consumers, so people are buying what they
like more and natural selection kills off the TVs in the population
set that are not selected. Part of that drive is, naturally, the need
to be entertained, but another part of it is that the more powerful
the display the more data it can deliver. This can and should be
harnessed by architects.
“When viewers have the ability to naturally interact with an environment, or are able to affect and be affected by environmental stimuli, they tend to become more immersed and engaged in that environment.” - pg. 29:2
There is an argument for critical distance - maintaining a
distance from a design being reviewed so that the design does
not influence the review itself. However, architecture cannot be
reduced to a set of images as it often is in design reviews. When
a film production team looks at a cut of a film they do so in a dark
room - much like the audience would view the film when it comes
out. Likewise in architecture, being able to experience a design
while it is being made like it would be experienced by its users
after it is built seems like a useful ability to have.
Fig. 2.17 One Approach for Creation of Images and Video for a Multiview Autostereoscopic 3D Display cover.
Fig. 2.18 Touchable 3D Video System cover.
3 4
Computer Games and Scientific Visualization
This article by Theresa-Marie Rhyne examines the use and
impact of video game technology in scientific visualization.
Relevant quotes in textual order:
“The market dynamics of computer game applications are thus influencing computer architectures historically associated with scientific visualization.” - pg. 42
While scientific visualization does not sound like it relates to
architectural visualization, one can make poignant comparisons.
Both are data-driven. Both are group-reviewed. Both develop
diagrammatic visual products. Both require iterative or prototype
design stages. Both are model-based, forgoing an exhaustive
translation of the entire product, instead focusing on a simplified
representation. If scientific visualization can learn from video
games, architecture can too.
“Shortcuts in the rendering software to produce a more engaging experience tor the user might work well in a game, but geologists using the same digital terrain data in a visual simulation of fault structures are unlikely to trust what they’re seeing or be able to apply it on a real-life scientific mission.” - pg. 42
A point against interactive visualization - sometimes
simplification of data renders it too unreliable. This works in
a purely scientific framework. However in architecture, the
simplification happens from an impossible ideal - no architectural
render has ever become reality. Ever. Thus simplifying from
a pretty picture to a less pretty picture but gaining real time
interaction works in architecture. At the same time, there are still
moments in design where data is crucial, but in those moments
making the design interactive in real time gains little for the
designer. At that point one has to be a little professional on when
to use a certain tool and when not to.
“Games now represent the leading force in the market for interactive consumer graphics. Not surprisingly, the graphics hardware vendors tend to anticipate the needs of game developers first, expecting scientific visualization requirements to be addressed in the process.” - pg. 43
Here is an interesting observation - hardware development
occurs for the lucrative business - video games - first, and the
data analysis, less popular, business, second, even though
the data analysis business should have a closer contact with
hardware development as they have more specific requirements
for hardware. This is to point out that architecture should still
piggy-back on something else when it comes to visualization and
interaction tools - until, or if ever, it is a powerful business, tools
will not be made for it. It will have to find them itself.
Component-Based Modeling of Complete Buildings
This research report by Luc Leblanc, Jocelyn Houle,
and Pierre Poulin examines another system for automatically
generating architecture. While this is not fully near my thesis, it
is important to be aware of what else computer technology is
capable of that architects have not harnessed yet.
The only relevant quote:
“Shape grammars constitute the state-of-the-art in procedural
Fig. 2.19 Computer Games and Scientific
Visualization cover.
Fig. 2.20 Component-Based Modeling of Complete Buildings
cover.
3 5
modeling of building exteriors, and have produced high-quality results. However, even though modeling building interiors and exteriors appears similar, shape grammars have not yet proven to be a good solution for modeling complete buildings. In fact, since their creation, only a small number of grammars, such as the palladian, have been produced for 2D floor plan generation, and better solutions have been provided by optimization techniques. Moreover, despite 10 years of development, shape grammars have seemingly yet to be used to model complete buildings. ” - pg. 87
While tools exist to parametrically generate exteriors, or
otherwise surfaces, those tools are not being applied to spaces or
are otherwise only being applied in a limited manner. Architects
spend too long marginalizing their own trailblazers - this report
claims over a decade has been spent on developing procedural
shape grammars, yet none of those years yielded a complete
procedural building. Is this an unimportant field in architecture?
Perhaps, but why has it been in development for so long, if so?
Exploring the Use of Ray Tracing for Future Games
This research report by Heiko Friedrich, Johannes Günther,
Andreas Dietrich, Michael Scherbaum, Hans-Peter Seidel, and
Philipp Slusallek introduces a software technique called ray
tracing and applies it to full virtual scene generation, including
shadows, reflection, refraction, caustics and other complex
effects. The report proposes that computers are now powerful
enough that this is possible at realistic hardware scales.
Relevant quotes in textual order:
“Computer games are the single most important force pushing the development of parallel, faster, and more capable hardware.” - pg. 41
One more reason to look to video games for cutting-edge
visualization in a field that is almost primarily...visual. Architects
can spend all the time they want making window schedules but
at the end of the day the product will be something that is seen.
“Some features of this engine are realistic glass with reflection and refraction, correct mirrors, per-pixel shadows, colored lights, fogging, and Bézier patches with high tessellation. All of these effects are simple to implement with rudimentary ray tracing techniques” - pg. 45
This quote is useful because, on the off chance that I attempt
to develop a visualization software, I know that it may not require
a high-end graphics engine with hundreds of shaders and visual
tricks - it all can be done with one system.
“Because ray tracing computes visibility and simulates lighting on the fly the pre-computed data structures needed for rasterization are unnecessary. Thus dynamic ray tracing would most likely allow for simulation-based games with fully dynamic environments as sketched above, leading to a new level of immersion and game experience.” - pg. 47
Here the technology of ray tracing is advertised on the fact
that, since it does not need pre-computation (like having to wait
for a render), it would provide the opportunity for immersive
interaction. This makes sense, as the faster the experience is
accessed from when it was designed the more responsive the
user would be as the conceptual thread in the mind would simply
continue from one medium to another.
Adding a Fourth Dimension to
Three Dimensional Virtual Spaces
The only relevant quote (on facing page):
Fig. 2.21 Exploring the Use of Ray Tracing for
Future Games cover.
Fig. 2.22 Adding a Fourth Dimension to Three Dimensional Virtual Spaces cover.
3 6
“This paper first outlines the capabilities of X3D to show buildings at different times or states. It then examines how temporal data can be stored within XML and combined with model data in the form of X3D. This data is then extracted and filtered on the client computer through the use of XML technologies. The way in which buildings can be displayed at different times or states along with associated descriptive text is demonstrated.” - pg. 164
The general gist of this research report, by Robina E.
Hetherington and John P. Scott, is the apparent simplicity of
encoding time data into a model on the pseudocode level. That
is, it is not fundamentally difficult to store temporal versions of
a design within the files of the design. This is significant because,
again, it is so simple for architects to use these tools, or to develop
them, that it boggles the mind that they have not used them yet,
or frown on their use. The ability to encode time data within the
design, separate from animation, could show clients, or a review
board, what the design would appear like during different times
of the year, which sounds like a powerful tool.
Service-Oriented Interactive 3D Visualization of
Massive 3D City Models on Thin Clients
This research report, by Dieter Hildebrandt, Jan Klimke,
Benjamin Hagedorn, and Jürgen Döllner, points out how
cumbersome specialized hardware and software can become. In a
system designed to visualize massive models of cities, specialized
hardware was developed with specialized software and an expert
was trained to operate all of that...just to make a moving picture
of a city. This is a point against the tendency with architects to
make tools that are highly specific to one purpose or, worse, one
project.
“Until today only “monolithic” geovisualization systems can cope with all these challenges of providing high-quality, interactive 3D visualization of massive 3D city models, but still have a number of limitations. Such systems typically consist of a workstation that is equipped with large storage and processing capabilities, as well as specialized rendering hardware and software, and is controlled by an expert who controls the virtual camera and decides which information to integrate into the visualization through a graphical user interface.” - pg. 1
Generally, tools need to be general. A hammer that works on
only one type of nail is not a very good hammer. A rendering setup
that only works during day scenes is not very useful in the large
scheme of things. Likewise, a system for interactively visualizing
designs should remain flexible so that all architects can use it.
“these systems mostly lack the emotional factor that is immanent to today’s presentation and interaction devices such as smartphones and tablets” - pg. 1
This is an aspect I have strangely ignored - the emotional
factor of being immersed in a design. There is zero emotion,
except despair, in an architectural review. Let the building speak
for itself, let it inspire, motivate, drive the review. Such are the
fruits of an interactive visualization system.
Fig. 2.23 Service-Oriented Interactive 3D Visualization of Massive 3D City Models on Thin
Clients cover.
3 7
On September 18th, I met with Thomas Cortina, Associate
Teaching Professor in Computer Science at the Gates-Hillman
Center. Below are important points from the meeting:
• Thomas mentioned a number of names I could pursue
for further inquiry: Jessica Hodgins, Kayvon Fatahalian (with
whom I eventually had an interview), both of whom work in
computer graphics, Alexey Efros, who is at Berkeley and works
with computational photography, and Guy Blelloch, who was
the lead on the design committee on the client side for the Gates
Center while it was being built. Some of these ended up being
unreachable.
• He also mentioned several libraries that I could look
into (and eventually did): the ACM (Association for Computing
Machinery) and SIGGRAPH, both of which could have articles and
research on graphics related to architecture.
• Yet a third line of inquiry he mentioned were the research
branches of large tech giants such as Microsoft, Google, IBM, and
Pixar, which often publish reports on cutting edge research and
technology.
All of these paths helped me develop my literary research.
I n T e R V I e w s a n D Re V I e w s
Fig. 2.25 Thomas Cortina.
Fig. 2.22 The College of Fine Arts compared to the Gates-Hillman Center at CMU. Both reflect the style of their age: the College of Fine Arts is rigid, uniform, and measured, while the Gates-Hillman Center is open, dynamic, and constantly
adapting.
3 8
Fig. 2.26 Kayvon Fatahalian.
Fig. 2.27 Near-exhaustive computation brought up during the interview.
On October 8th, I met with Kayvon Fatahalian, Assistant
Professor of Computer Science in the Smith Hall. Below are
important points from the meeting:
• Simple lighting can be done up to any arbitrary geometric
complexity, but baking complex shadows becomes tricky, and is
the area where graphics systems start taking shortcuts.
• One aspect of thesis is making this statement: “I believe
it is possible...” Where are the situations where existing tools do
not meet the needs of architects; what is not good enough?
• If I asked about what architects want, the deliverable
would be a proposed solution. Conducting a survey of the efficacy
of visualization software in the field would be fruitful.
• With an interactive render versus a static one, there is an
aesthetic trade-off - the first looks worse, the second looks very
good. What particular things do architects want to do?
• The idea of how pre-rendered videos can account for every
possible virtual scenario. That, or a mix of pre-rendered and real
time. How does that apply to architecture?
The biggest points I got from this meeting was to ask myself
how would an architect approach such software and what would
they need of it. This allowed me to move forward with software
analysis.
3 9
During the first poster session, on September 18th, I got
feedback from various professors in the School of Architecture
as well as my advisors and other students. Below are points from
that feedback:
• A feasibility analysis would be useful, in the form of a
flowchart with yes/no pathways that would narrow down the
nature of the thesis. This idea I later incorporated into both the
Mind Map and the software flowchart.
• The architectural design process was suggested to be
important to keep in mind. The problem had to be framed both
from the point of view of the client (what does the client want
to see?) and from the architect (what does the architect want to
show?).
From the first poster session I got ideas on what my
midreview should include to explain and ground my thesis.
Fig. 2.28 Poster #1 shown at the first poster session.
4 0
Fig. 2.29 The midreview plot.
Fig. 2.30 The midreview brochure, showing both the outside and the inside.
The midreview, on October 21st, was when the greater ideas
of how I was presenting my thesis came into play. The plot’s color
scheme was designed as if one were staring at the world with
one’s eyes closed. There were also brochures and my website
available for perusal, which made its official debut on that day.
The midreview had the following feedback:
•What is the dimensionality of inquiry? What is too
interactive? What is not visual enough? Where is this on a scale of
realism to representation to abstraction? This pushes the nature
of belief.
•Every tool changes the field. Speculate on what this will kill.
Find how it will negatively impact architectural practice.
•In 1994 renders were made with 600 kHz processors that
mimicked hand drawings. At some further point, firms began
experimenting with realistic renderings, with no technical
expertise.
•Is technology pushed just so it can wow someone?
Anything with technology or design has this eventuality, but is
that the point?
•There is a caveat - that I am not a technical designer.
•Lastly, comments were made to the effect of “this is a
thesis. Where is your project?”
4 1
The second poster session, on October 25th, was the same
week as the midreview so it featured little development from
the work at the midreview. It was more of a ‘coming attractions’
setup. As such I had a projector with a video setup in front of my
poster showing a glimpse of things to come.
The feedback from the first semester midreview and the
second poster session, due to its positivity, allowed me to
continue in full swing with the software evaluations. However, I
knew that, for many, getting a basic understanding of my thesis
was important, and I had to focus on that as well.
Fig. 2.31 Highlights from the second poster session. The top right image shows the setup with projector.
Fig. 2.32 QR code for gif animations of
the poster.
4 2
The first semester final review was on December 8th.
The review panel provided a number of new and interesting
perspectives that I can use to move forward with my thesis:
•I need to address how architects will use this, especially with
BIM and delivering construction documents. My assumptions are
far above the set of common assumptions of architects. I need
to bridge this gap. I need to look again at other firms doing this -
consider why animation is paid for, not in-house.
•With video games, there are other aspects than the visuals
that can benefit architects, like pathing, AI simulation, etc.
•When are beautiful sketches used compared to the GRID?
Is it detrimental to show this to a client, since they won’t use their
imagination anymore? Different audiences will use it differently.
•Different levels of information can be shown - maybe
abstraction is a tool architects want: the GRID can still have
motion, but does not have to be photorealistic.
•Video games and films are made to be mass produced, very
unlike architecture, and consider the social aspects.
•This exists, so what is the question? Will it eventually
become mainstream? Address the trend, and why accelerate it.
•What is a tutorial? Develop demonstrations, show not why,
but how - prove by example.
Fig. 2.33 The full final review plot, not including the projected video.
Fig. 2.34 The projector and speaker setup in front of the final review plot. The projector was used to project a moving graphic and a video.
4 3
The second half of the semester focused on engaging
software research with the literary research I did during the first
half of the semester. That involved an extensive analysis of various
software packages. These software packages are outlined in the
following pages. The analysis will follow the same thorough path
outlined on the facing page.
The main purpose of this part of my thesis is, within the
general context that my literary research created, to find a place
for visualization software in architectural practice. This is a two-
pronged development: the first prong is to actually find a capable
software package that can perform baseline photo-realistic
rendering and is flexible enough for a variety of applications.
The second prong is to approach the problem from the side of
architects: if one of these software is capable of these basic tasks,
what advanced, architecture-specific, techniques should they be
able to do? For example, should this software be able to simulate
people mingling in a project? Water collecting on roofs after a
heavy rain? Structural fatigue?
so f T w a R e Re s e a R C H
4 4
Fig. 2.35 Software research path.
4 5
The software I reviewed were Octane as a plugin for
Rhinoceros, Vray RT, which is part of Vray, Arauna2, a separate
program, UDK and CryEngine, which are video game software
suites, Blender Cycles and LuxRender, both experimental and
the first built into Blender, Unity3D, a video game development
suite, and Lumion, which was made specifically for architectural
visualizations. On the left are comparisons for each software in
each category on a scale of 0 to 10, subjectively.
I thoroughly analyzed each software for its pros, useful
features and benefits, its cons, where the software was hard
to use or had drawbacks, its software context, how it related
to a default installation of Rhinoceros and Vray, its rendering
features, what kind of rendering effects it could do, its rendering
drawbacks, what kind of shortcuts did it take to achieve real time
rendering, and its delay load, how much more time it would take
to work with this software compared to a render in Vray.
After considering everything, I found that none of the
software achieved high points in all categories. The choices I think
I have are Arauna2, Octane, CryEngine, and Lumion. Ultimately
it will be either Octane, given an interactive walking script is
made for Rhinoceros, or CryEngine, if I can streamline its import
process. Arauna2 would be nice, but it is still in development.
Lumion is almost there, but has too many interactive drawbacks
and does not appear to support scripts.Fig. 2.36 Summary of the software evaluations.
4 6
Octane sets up very quickly once loaded in Rhinoceros. The
default values are very good for an average Rhinoceros model.
The controls and materials are easy to define. It also can sync
with Rhinoceros’ camera. While it includes its own sunlight and
sky system, like Vray, it is built in, and needs to be reconfigured if
the scene already has a sun light. There are a lot of options - but
not all of them have much visible effect. It is GPU based, so other
programs are not heavily affected.
The biggest drawback is the renderer itself - path tracing
appears very fuzzy until the camera stops, after which the view
resolves within seconds. The camera can be set to only show the
view after a few samples have been calculated. The rendering
quality is fixed, so if it is slow then it will always be slow. Scene
complexity does affect it somewhat. Also, the viewport needs to
be updated when new geometry is created. Other cons are that
lights have to be set up as emitter surfaces and it does not appear
to use bump maps to simulate detail.
Otherwise, it can do all material types, depth of field, and has
advanced camera controls: exposure, ISO, gamma, saturation,
etc., and it can be networked.
The delay load is marginal. Time might be spent on setting
up materials, converting lights to emissive surfaces or trying to
find features of Vray that are not present in Octane - such as the
different renderers, animation controls, camera types, etc.
Fig. 2.37 Octane render logo.
Fig. 2.38 Snapshots of Octane’s controls and render viewport in Rhinoceros. Clockwise from top left: Basic scene featuring sunlight and sky modeling, depth of field, and reflections; Another example of an imported scene, featuring materials; Complex scene rendered rapidly; Scene with millions of triangles with minimal mesh conversion into Octane; Comparison with Vray RT, with similar materials; Comparison with the regular Vray using the sample scene I created, and lighting
matched as closely as possible.
4 7
The workflow in Lumion, which is separate from Rhinoceros,
is rapid and configurable. It definitely seems to come from a video
game background, as it has easy quality controls (compared to
CAD software, where preview controls are hard to access).
The import process is fast and intuitive, with a large library of
models of people, trees, and objects. It features terrain sculpting
and water bodies, with an ocean that has configurable waves.
However, the full version is not free.
The biggest drawback is that the aim is for pre-rendered
videos and images only. There is no walking mode and the
camera is a standard flying camera, though it can switch to orbit
via a button press. Below the ‘high quality’ setting, the rendering
looks very cheap. There are only a few fixed cloud arrangements,
though that is understandable given the task of photographing
a variety of clouds. There is a compromising feature though -
the clouds can be adjusted in density (which seems to have no
effect on the sun, and the clouds do not cast shadows). The water
customization is nice, but it is fairly fixed in style.
Otherwise, models can have any materials, but refraction is
by normal map only. Since it imports .obj files well, UV mapping
can be done in Rhinoceros.
Using it takes only several minutes. A scene can be set up
with the library of objects quickly, and the camera and UI controls
are fairly intuitive.
Fig. 2.39 Snapshots of Lumion’s controls and viewport. The menus are all flyout, meaning that once a scene is loaded it takes up the whole screen except when a menu is opened. Clockwise from top left: The sample scene with approximate shaders, note how water was used to approximate refraction; the same scene with materials and higher quality shadows, this was a performance hit on my laptop; The scene with the packaged elements included - a tree and a man, both affected by
light and animated, though the man walks in place.
Fig. 2.40 Lumion logo.
4 8
Arauna2 is a new experimental rendered that recently
revealed an evaluation version. So far it has many features: full
material support, including refractive, reflective, and specular,
lights support, built-in post processing and full screen filtering,
and a fixed sun model. It has a very easy to use UI, though the
camera controls are somewhat unintuitive. Another useful feature
it has that is rare to see are various extra rendering modes, such
as normals, depth, pure GI, rendering cost, and others.
Aside from the lack of a walking camera, the only drawback
is that it is still in development - there is no way to test how well
it imports models, or if it will have any more advanced features.
The camera does not collide with anything, but one can assume
there will be some way to use model collision. The evaluation
version uses a Unity scene as data, but that may be temporary. It
is also unknown if it will even be released as a separate program
for visualization - perhaps it will only be licensed for video game
developers. It does use path tracing, which is as always grainy
during motion. One minor point is that lights had hard shadows.
The delay load is unknown, but most likely marginal to
fractions of an hour, depending on the import process. This
renderer is very promising.
Fig. 2.42 Snapshots of Arauna2’s controls and viewport. The menus are all overlays, meaning that once a scene is loaded it takes up the whole screen except where a menu is, and everything can be hidden via a button. Clockwise from top left: Pure GI shading with depth focus in the back; Path tracing with focus in the front showing light effects and simple specular; Example of full scene reflection, which had no impact on performance; Example of refraction, some caustics, and
customizable light.
Fig. 2.41 Arauna2 logo.
4 9
Vray RT is the narrowest transition from regular Vray use,
though it lacks many features that the other renders have. Its
main draw it that, simply, it is a different button to press to do a
Vray render.
It appears to be a reduced renderer, and does not approach
Vray’s usual quality, thus seeming to be only for preview purposes.
Otherwise, ray-traced shadows and materials are rendered
accurately. The sun and lights are still processed properly. The
camera can be synced to Rhinoceros’ camera and does not
feature any other camera controls, like walking.
However, compared to more focused efforts like Octane or
Arauna2, it is grainy and resolves fairly slowly.
The delay load is minimal. It is only a different button away
from a regular Vray render. If nothing else can be done or used, it
is an available alternative.
Fig. 2.43 Snapshots of Vray RT’s viewport in Rhinoceros. Top to bottom: The render viewport by itself; The renderer, left, compared to Octane.
Fig. 2.44 Vray logo.
5 0
UDK (Unreal Development Kit) is a free software package
specifically made to develop video games. It is a large download
(1.9 GB) that features an extensive library of models and other
elements that can populate a scene and several rapid template
setups with preset sky and sun arrangements.
The import process must use Blender to convert .obj files to
a file format for UDK, .ase. Then, shadowmaps need to be baked
fairly quickly fast, but must be done again after any change.
Materials have to be set within UDK and are limited to simple
shaders. Sky and sun can be changed, and UDK has various types
of lights. Collision is a matter of a toggle.
The biggest drawback is that mesh import glitches at
65535 triangles, limiting the detail of complex models, requiring
them to be split into several chunks. It also takes around five
minutes to start. Many features in UDK are totally unnecessary
for the visualization itself. The sun light does not interact with
the atmosphere, requiring manual adjustment. Lastly, UDK uses
vertex lighting, causing shadows to appear off or inaccurate.
Otherwise, UDK has interactive walking. The camera bobs to
the motion of moving legs and there is a slight motion blur.
The delay load can be fractions of an hour, depending on any
issues with the import and basic materials exist or can be found.
Fig. 2.46 Snapshots of UDK’s controls and viewport. The viewport functions just like the one in Rhinoceros, where wireframe orthographic views can be set up. Clockwise from top left: The raw scene import with basic shadows calculated; The same scene with materials applied from the included library; The content browser, which shows the materials, objects, and other elements that come with
the software.
Fig. 2.45 UDK logo.
5 1
CryEngine is another software suite for making video games.
Even though it is newer than UDK it runs fairly smoothly (~20 fps)
on low-end systems. It also comes with a large library of models
that can populate a scene, like trees and rocks.
The huge drawback I experienced was that the export
process is long and arduous and requires either Blender
(unofficially) or 3DSMax (or Maya). The export process requires
significant set up in Blender. 3DSMax export is faster, except
material definition is faulty. In both very specific steps need to be
taken, with nearly any step prone to glitches, and a slip anywhere
may mean improperly assigned materials or a lack of collision.
However, once the meshes are imported it is fairly simple to
set up a scene, especially with a template file. All material effects
can be simulated with shaders, the sky and sun are realistically
modeled and an ocean or bodies of water can be made. There
is interactive walking just like in UDK, with the addition of the
walker’s shadow. The shadowmaps are entirely dynamically
generated and approximate GI. Lighting is simple but effective.
It may take multiples of an hour to bring a scene into
CryEngine from Rhinoceros. Even with practice there is a lot of
preparation that has to happen and not all of it is intuitive.
Fig. 2.47 Snapshots of CryEngine’s controls and viewport. The viewport only shows a 3D view of the scene, concordant with WYSIWYG. Clockwise from top left: The sample scene imported without any materials, featuring real time shadowing, sun, and sky; The same scene being tested, with materials, a shadow from the viewer, and the sky altered due to a lower sun angle; A view of the 3DSMax import
pipeline, where materials are assigned.
Fig. 2.48 CryEngine logo.
5 2
Blender comes with an experimental path tracing renderer
called Cycles. It has very few controls, which replace Blender’s
default controls once it is activated, thus there is less to learn
of the actual renderer once one has knowledge of how Blender
works. The path tracing rendering is very fast - the scene resolves
to an acceptable quality within seconds if the camera is still. Also,
since Blender is free Cycles is free as well. This also means there is
a large DIY community of graphic modelers and designers.
Cycles supports Blender’s light objects and material
definitions, with many presets including reflective, refractive,
cartoon, and others. While moving the view is pixelated but is not
choppy, which is a better solution than that used in Octane.
The biggest drawback is that it requires some knowledge of
Blender, which has a steep learning curve. If geometry is imported
from an .obj file, materials have to be reassigned. Blender’s sun,
as it is handled by Cycles, does not have sunlight modeling - it is
just a distant light at a given angle, though a modeled sky can be
set up. Blender does not easily support walking.
The delay load is fractions of an hour or more - added to how
much time it would take to learn Blender, setting up a project here
compared to Vray takes more effort, including changing mouse
controls, changing how objects are placed and moved, and more.
Fig. 2.50 Snapshots of Blender’s controls and render viewport. The viewports in Blender can be variously configured. Clockwise from top left: The sample scene with soft shadows and full materials; The same scene with harder shadows; The
scene as it resolves with one sampling, showing the graininess it begins with.
Fig. 2.49 Blender’s logo. Cycles does not
have a logo.
5 3
Unity is similar to both the game suites and to Blender in that
it is designed to make games but has its own modeling tools. Its UI
is relatively straightforward. The free version has many features,
enough to do basic visualizations. While it can import .obj directly,
Blender may be required for additional material or UV setup.
A big drawback is that many advanced features present
in the other software are not included in the free version and
the features that are present are fairly weak in quality. The sun
and sky need to be faked to achieve various daytime lighting
situations and the shadows seem to be fairly low quality and need
to be calculated, a process that takes several minutes.
Otherwise it has material shaders, some kind of real time
shadows and supports light objects. Walking is supported after
some setup. Mesh collision can be easily set.
The delay load is fairly small - fractions of an hour - any
extra setup in Blender and importing assets into Unity take time,
although template scenes may be possible.
Fig. 2.51 Snapshots of Unity’s controls and render viewport. The viewport switches to game mode when the walking is activated. Top to bottom: The sample scene with some basic materials showing dynamic shadows; Precomputed
shadows, but at a low quality.
Fig. 2.52 Unity logo.
5 4
LuxRender is fairly fast and comes with a material library,
but it does not provide any interactivity. It is a step backwards,
using new software rendering but not using it advantageously.
It is a plugin renderer for Blender and works on the same level as
Cycles. It likewise changes various settings and generates a new
viewport when the render is started.
It renders a frame at a time, like Vray and due to the new
viewport it is difficult to move back and forth between the design
window and the render window.
Its delay load can get to fractions of an hour. Material
settings and assignments are nothing like those of Vray and are
somewhat clunky, on top of learning the workflow of Blender.Fig. 2.54 Snapshot of LuxRender’s controls and render viewport. There are more controls in Blender’s menus. This is highly similar to Vray’s viewport.
Fig. 2.53 LuxRender logo.
5 5
On November 18th I received a new graphics card I purchased
a few days earlier. This card was a Nvidia GeForce GTX 660 Ti,
replacing an ATI HD 5770, and the reason was purely because it
had hardware that enabled the use of or the faster application of
several of the software packages I looked into. Nvidia graphics
processors (GPUs) have a technology called CUDA that uses
parallel processing to do graphics tasks. The software that uses
this technology, Octane, Arauna2, and other path tracers, would
not otherwise work with the ATI card that I had before. I was
able to use Octane at reduced settings on my laptop, as it had an
Nvidia card albeit one of lesser quality, but the others would not
work with that card because it was too old.
The laptop card was a GeForce 130M with compute capability
(a property of CUDA technology) of 1.1, whereas the 660 Ti, by
comparison, has one of 3.0. The laptop card also has only 32 CUDA
cores, whereas the new desktop card has 1344. Also, for the video
game engines, the new card is roughly 50% stronger than the ATI
card I had before, so I can push those engines further to achieve
higher quality visualizations.
Buying the new card (a $259.99 value) was the best option
for my thesis in terms of hardware because it was readily available,
Ha R D w a R e Re s e a R C H
Fig. 2.57 The old ATI card.
Fig. 2.55 The new card, left, versus the old card, right.
Fig. 2.56 The new card inside the desktop tower.
5 6
enabled the use of software for my thesis, and demonstrated that
my thesis can exist without expensive or cutting edge hardware
like virtual reality headsets, new means of interaction like the
hardware Adobe is developing, or immersive room-sized display
setups.
The initial limitation of the low-end hardware on my laptop
and the unusable hardware on my desktop still played an
important part in my thesis because it showed that this software
could be used on existing, potentially old, hardware, though with
severe drawbacks and shortcuts.
Fig. 2.58 The Oculus Rift virtual reality headset in action. This is an example of unattainable
hardware.
Fig. 2.60 Unboxing the new card. It came with an instruction manual, a drivers disc, and extra cables. The card was distributed by ASUS, which also added the
cooling system.
Fig. 2.59 Adobe Mighty and Napoleon. Mighty is the triangular pen, Napoleon is the ruler.
5 7
Fig. 3.1 Serious Editor 3.5 by Croteam. This kind of software is used by video game developers to create virtual worlds - much like architects do with CAD software,
except with materiality and lighting as part of the toolset.
Fig. 3.3 Unreal 4 by Epic Games. This is a future engine currently in development that, while it still uses shaders, simple lighting, and other standard methods, pushes
them to their limits to achieve photorealism.
Fig. 3.2 CryEngine Sandbox by Crytek. This is a much more recent video game engine and favors dynamic shadow generation over the use of pre-computed
shadowmaps.
Fig. 3.4 Luminous Engine by Konami. This is also a future engine currently in development. Engines like this are at the forefront of video game engine technology,
pushing what is possible with shaders and graphics software.
D e l I V e R a b l e s
aP P l I C a T I o n s
5 8
Fig. 3.5 Help files and documentation for various graphics software. Clockwise from top left: Unity; Blender; UDK; Rhinoceros. These range in quality and depth, with some featuring text and image descriptions and others even including video. Unity was the only one that read from an included file, the others either embedded or opened a browser
page to an online database.
5 9
Fig. 3.6 Fallingwater in Half-Life 2 by Kasperg. This is a demonstration of modeling a real building in a video game environment.
Fig. 3.8 City scenes in Brigade 3 by Otoy. This is the cutting edge of cutting edge path tracers.
Fig. 3.9 Fox Engine by Konami. One of the images in each set is the engine, the other is a comparative real life photograph. Which images are the engine?
Fig. 3.10 Euclideon Engine. This uses a method I did not explore - voxels - as it is more about generating geometry rather than photorealism.
Fig. 3.7 House in UDK by Luigi Russo. This student project, modeled in video game software, showed that the same goals that students use CAD
software for can be applied to video game engines.
6 0
Fig. 3.11 Path tracing method, sample images. This shows an exhaustively detailed physical environment rendered with full lighting and materiality at interactive speeds. On the right, water effects are also simulated.
Fig. 3.12 Las Vegas Bellagio Comparison in CryEngine by IMAGTP. This is a photo-realistic demonstration of a real building compared to a photograph taken at the same location.
6 1
Depending on which software I move forward with, the
next steps of my project will be either lightweight coding or
heavyweight streamlining or coding.
The Octane approach assumes that Octane is set up within
Rhinoceros and the only thing missing is an interactive control.
The range and nature of this control will vary, as simple horizontal
camera control by forward impetus and turning is much simpler
than also having camera bob, gravity, or collision detection.
The CryEngine approach assumes that it is installed and a
Rhinoceros project is available and the only thing in the way is the
cumbersome and complex import process. The range and nature
of streamlining this process will vary from simply documenting
comprehensively and cleanly how to do it with the least mistakes,
to attempting to enhance the plug-ins already existing to attempt
to automate the process further.
Once one of the above is in place, the next steps are more or
less identical. After both real time rendering and interaction are
achieved I need to document further features such as material
assignment, any way for collaboration or portability, streamlining
controls, general use principles or shortcuts, and the like.
mo V I n G fo R w a R D - so f T w a R e Pa C k a G e
Fig. 3.13 The Octane approach, where only an interactive script needs to be made.
Fig. 3.14 The CryEngine approach, where the import path needs to be streamlined.
6 2
After that, I would attempt to compile a software package.
With Octane, that would involve everything but the software itself,
as it is not free and would need to be purchased. Otherwise, there
would be a zip file, or even a small self-installer that will consist
of plug-ins, help documents, videos, and so on. With CryEngine,
the package will be more robust, as, at least theoretically, it may
include all of CryEngine, which is hefty at 1.9 GB. Since there
may be licensing issues I may also require that it be downloaded
separately, but as it is free that is less of an issue.
The software package, by its very existence, would be the
proof of concept for my thesis. However, at this time it is of a very
vague nature since there are too many variables in how I would
approach developing it. Workload-wise, developing the help files
and tutorials alone is a lot of documentation, and if I choose to do
some sort of scripting I would need time to familiarize myself with
the scripting languages that I would need to use.
Also, knowing my audience will be very important. As
different users will use it differently, I will need to frame it as such.
For an architect looking to use it as a design tool, to rapidly view
a project interactively with progressive visuals, it will be one thing
and have a certain feature set. For a client wishing to explore
a realistic simulation of a project, it will be another thing. For a
contractor wishing to see the assembly of certain elements it will
be yet something else.
Fig. 3.15 The breakdown of the software package.
6 3
mo V I n G fo R w a R D - be n e f I T s a n D De a T H
The software package foresees potentially great benefits
in the field of architecture. To understand where these benefits
come from, it is useful to review in a nutshell. THE GRID is a tool
meant to preclude the physicality of an architectural design via
software and hardware that is currently available and immerse,
by visual and other means, the client or architect in the design
before it is built. Since architecture is experienced through both
time and space, it is necessary that such a tool exist during some
early stage of design before the design is finalized and converted
into construction documents. That conversion is generally
done with BIM, as BIM is accurate and collaborative. Once the
BIM phase starts there need for the GRID lessens, as it can be
assumed the design will not change dramatically at that point.
Using construction documents, the architect and contractor
collaborate to produce a built space, too late to make many
changes. In the current design process, the real space and the
real time are only reachable in snapshots or animations generated
beforehand (animations could be understood as simply a series
of snapshots). The problem with this is that, due to the way one
experiences a still image versus a physical space, there will always
be experiential differences between what the design was before
it was built and what the design becomes after the construction.
6 4
Benefit #1 - Reducing the Gap: The first benefit of the tool is
that the experiential gap is narrowed or even removed. With the
ability to see a building through the medium of a computer screen
with realistic shadows, movement, light, and materiality, both the
client and the architect are brought to the same level. The client
probably has little experience working with CAD models and
renders, or animations, and lacks the preparation that the CAD
model, and working with that model, gives the architect.
What the client lacks is an understanding of the space.
This can be done by teaching the client how to understand
architectural orthographics, which is arduous and exacerbates
the problem (by reducing experience rather than expanding it)
or the client can have what they already understand, a visual
substitute for the real thing. Sketches work, but ultimately the
building will be something real, and this reality has to somehow
manifest early on.
Benefit #2 - Personalization: This allows the client to make
the project their own. By using computer interfaces they can
inhabit and explore the project. The power of the simulation is
that it uses the user’s own brain to their advantage by letting it
translate the motion within the virtual world to motion of their
physical self. The project becomes familiar and understandable.
This assumes that the client was not already swayed with
beautiful sketches, or other abstract representations of the
Fig. 3.16 The experiential gap between a still image and the physical space.
Fig. 3.17 THE GRID allows a viewer to experientially inhabit the project.
6 5
project. Where these other representations used imagination to
create the space, the GRID will use it to explore the space.
Benefit #3 - Prototyping: Even before the client uses this
tool, the architect her- or himself can use it to rapidly prototype
the experience rather than the assembly or the totality alone.
Architecture is far to big to be prototyped in full, and prototyping
little chunks only goes so far.
One can approach this by breaking down what architecture
is - the memory of time and space. Memory is the passage of
experience, time is a series of moments, and space is moment
given shape. Space is the easiest to prototype because an
architect can build a scale model - this will provide a sense of
the space. Time is also easy to prototype, because the architect
need only to hold the model for a while. Memory is a little harder
because the brain is smart - it knows the model is just a small
object. The architect needs to trick her or his brain, to get down
near the model and pretend it was big and thus come close to a
memory.
THE GRID does that and goes further. It also has the architect
make a model, and spend time with it, and get down close to it.
But it goes beyond - the architect can walk inside the model, the
architect can change the lights and the time of day, the architect
can flood a room with water or place other people in the space.
True creativity can flower then, when memory is achieved.
Fig. 3.18 Architecture is composed of the memory of time and space. Space can be prototyped with models, time can be prototyped by simply being around and examining the models, but memory is harder because it involves tricking the brain.
6 6
+ + + + + + + + + +
The thesis would also be malicious. As a tool it would
contend with orthographics, renders, and physical models.
Before construction documents and after things that can be
called sketches these tools have come to be standard in the
design pipeline.
Death #1 - Orthographics: With orthographics, should the
client be enamored with the GRID, they may not be interested
in plans or sections, even though those tools still provide
valuable insight into the spacial organization of a project and the
interaction of the systems within or between the spaces. Likewise,
if an architect is using an axonometric diagram to explain the
order in a project but the client does not see that order in the
GRID, the client may put less faith in the work the architect put
into the diagrams, demanding instead, perhaps unrealistically,
that the diagrams match the experience found on the GRID. In
an in-firm review the orthographics may be quickly cast aside
as experiential conversations are brought up only visible on the
GRID, raising questions as to why the architect spent time on the
orthographics over working on the GRID.
Death #2 - Renders: With renders, the overlap is sharper.
Given a regular pretty render and the GRID, the client may wonder
why the architect bothered to take one picture when the GRID
allows them to move around and take any and all the pictures
Fig. 3.19 A client may not care about orthographics if the GRID is compelling.
Fig. 3.20 Two architects, one waiting on a traditional render, the other already being group reviewed.
6 7
that they want, from any attainable angle. Back at the firm, the
architect is spending many hours working on a few renders while
another architect, working on the same project, in the same time
finalized the GRID, rapidly creating countless renders and videos
of the same project, all at an even high quality.
Death #3 - Physical Models: Even physical models may feel
the heat - much as with a render, one architect spends the whole
night crafting a model while another has created the GRID, with
full materiality, realistic sun shading, water bodies, and more. The
only difference is, the physical model is twirled in the hands while
the GRID is controlled by a keyboard and mouse. Even assuming
advanced hardware exists, the physical model is 3D printed with
full geometric detail...and the GRID architect uses an Oculus Rift
to create a virtual 3D display that delivers a near-real experience,
complete with depth information.
Sketching too may be impacted, though not killed - imagine
a precedent study being not just looking at photos and drawings
but exploring the GRID version of that building, perhaps modeled
with LiDAR, and documenting the experience. Perhaps one step
of design is quickly molding spaces on the GRID and experiencing
them for inspiration.
Construction documents can be reinforced by the GRID. An
architect can show a polished GRID on the construction site to
the team, showing how the project would look like, materials,
Fig. 3.21 Prototyping small pieces of architecture with physical models takes time and does not give an accurate rendition of the built end result. With a digital
substitute, the entire project can be prototyped and reviewed.
6 8
shading, landscape elements and all, as one moved from one
end of the building to the other. Since the GRID is intuitively
understood, the contractor would not need to learn a new means
of communication with the architect.
Also during construction, the architect, perhaps if she or he
now sends a floor plan to the tenements of a future apartment
building so that they can mock up their furniture arrangements,
can send the tenements the GRID, which they can use to explore
and make their choices in a medium they can understand. No
more need for the billboard proclaiming a future building - just
go on the website of the firm and download that building’s GRID.
Fig. 3.22 A client customizing a house using a real time visualization to get the exact appearance they want.
6 9
mo V I n G fo R w a R D -Im a G I n a T I o n a n D ex P e R I e n C e
The thesis needs to find its audience, for its audience
does not know the show is on. There are certain assumptions
inherent in the GRID that are so far away from the common set
of assumptions of architects and their connected fields that
I attempted to break down, and I isolated a few and tried to
address them, but many remain.
One aspect that I overshadowed was the reality that the
software implementation that my thesis is exploring is already
present in some arenas - some firms have used this as a design tool
and delivered it to clients as such. These unsung firms, however,
do not themselves see the benefits of spreading this knowledge
to the rest of the field. Perhaps because this is because they feel
entitled to uniqueness, perhaps they do not see results or believe
this delivery is more work than it is worth. Perhaps every person
they use it with has desired different things from it.
Different audiences will indeed react differently to the GRID.
Well-entrenched firms will not allow for yet another software
into their pipeline, while more open firms will see it as a design
tool, perhaps devaluing the photorealism for the interaction and
layered data sets.
7 0
The data sets firms may choose could differ from the general
one I focused on - that of photorealistic interaction by walking
in real time. Some firms, or even the client and maybe the
contractor may desire to explore the project while only focusing
on the hierarchy of spaces, or perhaps while emphasizing the
structure behind the walls. Their imagination could then be
guided depending on the type of communication.
The imagination of the recipient, be it client, contractor,
or fellow architect, would nevertheless still be engaged. While
abstract sketches or diagrams can communicate, nothing yet
gives the user the element of choice, the choice of experience,
over memory. The choice of what can be done, over what has
been done.
Would this lead to a death of the outdated cultural belief
that architectural products are drawings, and instead herald an
age where people see architects embracing the digital? What if an
architect wanted to do something other than what his profession
had intended for him or her? What if an architect dreamed of
something more, some means of taking their understanding and
making it the understanding of others?
THE GRID will give architects an ideal to strive towards. They
will still render, still make animations, still rely on CAD. But in time,
they will learn to use it, to make it shine as the sun. In time, it will
help them accomplish wonders.
7 3 Fig. 4.1 Formation of THE GRID logo.
7 4
Table of Contents 74
Introduction 75
Unresolved Issues 75
Initial Course of Action 77
Underlying Assumptions and Argument 81
The Tutorials 85
Overview of the Visualization Tutorials 85
Tutorials 1 - 8: Visualization 86
Overview of the Interaction Tutorials 114
Tutorials 9 - 12: Interaction 115
The Templates 125
Being on THE GRID 127
Difficulties of The Tool 127
Spring 2013 Project 129
The Design Challenges 131
User Interactions 133
Part two - table of Contents
I n t e r a C t I o n a n d P o s s I b I l I t y
7 5
THE GRIU as it stood at the end of the first semester
had three significant unresolved issues. These issues
were as follows:
1. Indecision Between Octane or CryEngine
This is a major issue because one approach, Octane,
is strongly photorealistic but lacks significant interaction.
The other approach, CryEngine, is strongly interactive but
is worse at rendering as it uses shaders and simplified
lighting. Thus I could not use one without a sacrifice of what
the other may provide. The option of using both existed but
then THE GRID could have become cumbersome with too
much new software to learn. Resolving this streamlined
my thesis.
2. How to Demo With a Lower-End Hardware Laptop
The laptop I had at the time was fairly old and, while
it could run both approaches fairly well, did not represent a
modern piece of hardware or the quality of hardware than
an architecture office would have. Buying a brand new
laptop was not an option due to cost, but if there was a way
to either rent a laptop or simulate better hardware (perhaps
Unresolved IssUes
I n t r o d U C t I o n
7 6
using my stronger desktop as proxy), then I would have
been able to demo more effectively. Resolving this was not
fully necessary yet would have been a great benefit.
3. Determining the Feasibility of Redistribution
Since both approaches involve a whole separate piece
of software my training materials would have required the
user to install one or the other piece of software. Octane
has a hefty price while CryEngine is free but is nearly 2 GB
in size. One or the other would have needed administrative
privileges on the target machine, but if the assumption was
made that the software were acquired separately then
the training materials could have remained valid. Since
everything already depended on Rhinoceros and/or Vray
being present on the target machine, the point may have
been moot on a technicality. However, on top of that there
was the issue of burning to a DVD or preparing an online
archive, both options that I figured out later. Resolving this
enhanced the practicality of my thesis.
Fig. 4.2 CryEngine intro splash.
7 7
The thesis, going into the second semester, can once
again be summarized in a one-sentence, three-sentence,
nine-sentence format. Note that this was the initial course
of the thesis for the second semester but as the work went
on the course was expanded to include student testing and
interaction.
1: To Educate and Advance the Field of Architecture
Using Existing Photorealistic Real Time Interactive Design
and Video Game Software.
3: The education would come in the form of complex
and user-oriented tutorials and instructions. Even though
the veracity of THE GRID was proved last semester, a more
concise summary of its importance would accompany the
training materials. Two approaches, one using a piece of
design software and the other using a piece of video game
software, would be the focus of the training materials and
serve to advance the field.
9: THE GRID is about maximizing tools already
available or on the cutting edge, as very few instruction
manuals, or any, exist for the tools in question. These tools
InItIal CoUrse of aCtIon
Fig. 4.3 The second semester focused on the space between THE GRID lines of the first semester: GRID cells instead of GRID lines.
7 8
focus on photorealistic real time interactive visualization of
digital models, assuming that there will always be a digital
model to work with in a modern project, as otherwise the
project will get nowhere. Of two approaches chosen last
semester, one uses a path tracing plugin for Rhinoceros
that is orders of magnitude faster than Vray, but as it is a
recent piece of software it has not gained popular traction
The second approach is a video game editor, CryEngine,
which provides significant interactivity but at reduced
rendering quality. For both or either, training materials will
have to be made so that a designer familiar with Rhinoceros
and the Vray workflow will be able to, with the help of the
training materials, use THE GRID to create photorealistic
interactive real time visualizations. The training materials
will go further: once the basics are set up, more advanced
techniques will be detailed. Once the user is making the
visualizations their benefits will become self-evident -
presence, interaction, and a sensory fulfillment combine
to create a prototype of the experience, an ultimatype. The
training materials will be packaged in a way that can be
redistributed. That is THE GRID.
[][][][][][][][][][]
The second semester therefore began with focusing
on testing Octane and CryEngine as valid options for THE
7 9
GRID. I planned to look for pros and cons of each system,
in regards to user experience: How many steps does it
take to go from a raw Rhinoceros model to an interactive,
real time, model with materials and lighting? How possible
is it to add dynamic elements, like moving people, cars, or
trees? How many individual pieces of software or plugins
are required? Following that, I planned to attempt to
quantitatively examine the output of each system to figure
out if using one, what does a user lose from the other, and
vice versa.
The next step would have been to pick one, or
both, and establish the nature of the training materials.
This meant I had to develop tutorial paths: make basic,
intermediate, and advanced tutorials sets. Each path could
expand in detail but not complexity, the tutorials had to be
easy to use, with quality videos and annotated diagrams
and screenshots. I would have had to do a hardware
spec analysis - not everyone knows what a GPU is, or
CUDA. Lastly, I would have had to determine methods
of distribution: digital - PDFs, YouTube, etc. or physical -
booklet, CDs or DVDs, etc.
However, as the work went on the course expanded
beyond that. The initial development of the tutorials as an
online resource was over about a third of the semester in,
8 0
so it was decided that I would continue with THE GRID
by opening it to fellow students. This meant creating an
environment on THE GRID and design challenges for the
students to perform so that the true possibilities of THE GRID
could be determined. A computer then had to be acquired
since my laptop had continued to degrade in hardware
quality and attempting to have the students use THE GRID
on their own machines was a failure. With a computer set
up and an environment made for the students, they were
able to test THE GRID.
Fig. 4.4 A brief analysis of the possible avenues for testers of THE GRID. The most likely source of testers was colleagues, and the least likely were online people.
8 1
It was important, before work after the first semester
could proceed, to determine any underlying assumptions I
had about the context of the thesis and reconsider my core
argument. The underlying assumptions of THE GRID are as
follows:
• Architects are in the business of making digital models
before converting them into construction documents.
• Architects’ use of digital models is far behind their
use in other fields, such as medical imaging, geological
surveying, simulations, and certain types of graphics-
heavy video games.
• The technology and software used in the School
of Architecture at Carnegie Mellon University is indicative
of technology and software in use in architecture offices
worldwide.
• Architects’ presentation of digital models is limited to
static renders and pre-configured animations.
• Architects have enough cash to consider upgrading
hardware (if needed) and software.
UnderlyIng assUmPtIons and argUment
8 2
• If rapid or real time photorealistic renderings are
effective in other fields, they can be effective in architecture.
• Presence exists.
• A digital model can be explored with a navigational
interface effecting better understanding of the project; this
is apart from looking at plans and diagrams.
• Static images cannot represent a project that is
meant to be physical.
• Architecture is digitally stagnant and both the public
and construction sectors are often misinformed as to the
nature of architectural design and education.
[][][][][][][][][][]
The core argument could be broken down into the
following:
• Architects need a way to experientially prototype the
final design in the digital realm.
• Architects - professionals and students interested
in orchestrating space and time with programmatic and
social implications. They deal primarily with visual and
tactile experiences, but sound also comes into play. They
delve into invisible properties, like the structure or wiring
8 3
in a building, only diagrammatically or by reference, i. e.
leaving reasonable room for those features so that they
can be developed when the project is converted into a
format that can be physically constructed at full scale.
• Need - architectural stagnation is brought on by
reliance on old tools, practices, and paradigms. Therefore,
a need arises to adopt more modern ideas.
• Way - tools, methods, repeatable and reliable
practices.
• Experientially - the experience is the overarching
theme to architecture - the sensory interface with space and
time leads to memory which is understood as experiences.
There is no way to experience a building before it is built,
but photorealistic digital tools combined with interactive
controls can be an effective surrogate.
• Prototype - to find out what it is like before it
exists. Think of cars, books, phones, and toys - all can
have a rough full scale prototype developed rapidly during
design. Architecture is too big to be fully prototyped, and
prototyping a piece is like only printing a page of a whole
book or only plastic-injecting a wheel of a toy car - it is
accurate, but limited.
• Final design - while design continues until the last
8 4
brick is in place, in a best case scenario the major motions
of the project are figured out before construction documents
are made. The final design is a 3D, fully digital model. This
is the case, as based off of the first underlying assumption.
• Digital realm - fully computer simulated environment
with physical assets (like textures) imported. This is
important because, as assumed, the digital realm has been
successfully used in other fields using advanced software
that is currently available. The digital is here to stay and
must be embraced.
8 5
t h e t U t o r I a l s
The CryEngine tutorials at whatisthegrid.tumblr.com
are the first step in analyzing how a user experiences THE
GRID. They are a sequence of twelve tutorials, instructions
for use, that guide someone only familiar with Rhinoceros,
Vray, and has access to 3dsMax, through using THE GRID’s
chosen software, CryEngine to arrive at a fully interactive
visualization of a chosen model of one of their projects.
Each tutorial outlines with text and image the basic steps
of what needs to happen and what can happen on THE
GRID.
The website and the presentation of the tutorials are
designed to convey a sense of fluidity, of breaking the rigid
grid lines and uncovering the cells within. Website elements
fluidly appear and shift, the mouse cursor produces a trail
of tiny grid cells, the tutorials themselves are laid out in a
flowing manner, and a sidebar available on each tutorial’s
page allows for direct travel from tutorial to tutorial. The
same website was supposed to eventually also contain
design challenges which would have been part of testing
THE GRID with potential users.
overvIew of the vIsUalIzatIon tUtorIals
Fig. 5.1 QR code for the tutorials.
8 6
1. InstallatIon
In this tutorial we will learn how to install CryEngine
and some tools for 3dsMax.
THE GRID requires CryEngine and 3dsMax on top
of the core modeling software, which is assumed to be
Rhinoceros. CryEngine is currently only available for a PC
machine, so switch to one if available. For this tutorial, it
is assumed that both 3dsMax and Rhinoceros are already
installed.
To install CryEngine, you must first set off some time
and have a good internet connection, as the download is
roughly 1.9 GB.
Visit cryengine.com. The home page should have a top
bar much like the one below:
Click on “GET CRYENGINE.” A new page appears as
below. Click on the “FREE SDK” monitor.
In the next screen, click on “DOWNLOAD NOW.” Above
this button you can review the “System Requirements” and
see if your machine can run CryEngine. CryEngine is a fairly
flexible system, however, and can handle many different
hardware setups.
At the last screen click “Download now” to initiate
the download. Note the size below - even at high internet
speeds this may take a while.
Once downloaded, extract the archive to C:\CryEngine\
1. InstallatIon
8 7
Bin64\ if your operating system is 64-bit or C:\CryEngine\
Bin32\ if your operating system is 32-bit using your favorite
file browser. Locate Editor.exe. Right-click this executable
and select ‘Copy’. Then, go to your desktop and right-click
anywhere and select ‘Paste shortcut.’
Now, navigate to C:\CryEngine\Tools\ (Or the
corresponding folder for an installation elsewhere). Right-
click CryToolsInstaller.exe and select ‘Run as Administrator’
with the shield next to it. Accept any notifications. A
window like the one below should appear.
Hit ‘Next.’ You can uncheck all but ‘Autodesk 3ds Max.’
Hit ‘Next’ again. The installation should be a success. Hit
‘Close.’ If Windows pops up a warning message hit ‘Cancel’
and ignore it. 3ds Max has now gained useful tools for
exporting to CryEngine.
In the next tutorial we will learn what needs to be done
within Rhinoceros to prepare it for export.
2. Setting Up RhinoceRoS
In this tutorial we will learn how to set up Rhinoceros
to properly handle exporting to 3dsMax.
There is actually very little setup that has to be done
in Rhinoceros. Most of the CryEngine-specific settings are
established in 3dsMax itself.
To start off, begin by opening Rhinoceros:
2. settIng UP rhInoCeros
8 8
Wait until it loads and presents you with an empty file.
The only thing to make sure of with your Rhinoceros
model is that everything that you want to have a material
has a material assigned to it either directly or by layer.
So, begin by opening a typical Rhinoceros model.
On the Properties tab on the right, select an object and
locate the ‘Material’ button. If the tab is not there, you can
type ‘Properties’ to bring it up. In the ‘Assign material by’
dropdown, note the various options. From here you would
either select ‘Object’ from that menu (‘Layer’ is the default)
or set the material by layer, which is done in the layer
panel by clicking the colored circle under the ‘Material’
column for each layer. Both choices would end in the same
window or tab. Above, third, is the appearance if you had
chosen to set the material by object. You can either start
modifying the default material, which will automatically
create a copy preserving your changes and naming it with
a default name, or you can press the ‘New...’ button to
create a new material.
8 9
Pressing that button will open the preceding window.
Hit ‘OK.’
Either way, you will end up with a material assigned
that is named ‘New material 001’ or similar. That is fairly
non-descriptive so a good idea is to rename it. It will be
helpful to name the material after its real-word equivalent,
as later on it will be easier to assign properties to it in
CryEngine when all you will have is the name of the
material (as object-material assignments can only be done
in Rhinoceros or 3dsMax). Right-click the material name
and select ‘Rename...” to give that material a name.
In the settings below, you can change the basic color
by clicking on the colored rectangle to the right of ‘Color’
in the ‘Basic Settings’ section or change the texture by
clicking on (empty - click to assign) in the ‘Textures’ section.
These settings and UV mapping are for more advanced
modeling. If you want to explore, the texture mapping can
be accessed by clicking the ‘Texture Mapping’ button at
the top.
In the next tutorial we will learn how to use CryEngine
and how to load a template.
3. Loading a TempLaTe
In this tutorial we will learn how to work in CryEngine
and how to load a template world prepared with a basic
3. loadIng a temPlate
9 0
scene into which models can be dropped.
Begin by launching CryEngine.
Here you will probably need to sign in to the CryDev
network. Go to crydev.net and click on ‘Register’ in the
upper right; this process is free and Crytek does not bother
you with emails or whatnot. Once you have credentials,
sign in and it will remember that computer forever.
You will be presented with the default layout and a
welcome box. Note that I reorganized the toolbars a bit to
optimize space. For now, let’s see what a blank level looks
like and why it is useful to work from a template. Click on
‘New Level’ on the welcome box to start with a blank level.
You will be presented with an options box. Give the
level any name and ignore the terrain settings and click
‘OK.’
You will be presented with another options box, this
time for the terrain. Keep it as it is and click ‘OK.’
Now you will be presented with a view into the virtual
world containing a large ocean, sky, and ground some
9 0
scene into which models can be dropped.
Begin by launching CryEngine.
Here you will probably need to sign in to the CryDev
network. Go to crydev.net and click on ‘Register’ in the
upper right; this process is free and Crytek does not bother
you with emails or whatnot. Once you have credentials,
sign in and it will remember that computer forever.
You will be presented with the default layout and a
welcome box. Note that I reorganized the toolbars a bit to
optimize space. For now, let’s see what a blank level looks
like and why it is useful to work from a template. Click on
‘New Level’ on the welcome box to start with a blank level.
You will be presented with an options box. Give the
level any name and ignore the terrain settings and click
‘OK.’
You will be presented with another options box, this
time for the terrain. Keep it as it is and click ‘OK.’
Now you will be presented with a view into the virtual
world containing a large ocean, sky, and ground some
9 1
distance under the water. The first thing to do is rearrange
some stuff (you may have noticed the previous views
looked different than default). The default layout is below:
To save some space, merely click and drag the two
lowest menus among all of the buttons in the top bar to
the side so it appears like it does in these views. Also,
feel free to drag out the bar on the far right out and close it.
Next, note that you may have the info overlay turned
on, as below. Click the ‘i’ in the upper right corner above
the viewport to cycle through the displays until it is off
or showing information that you find useful. This overlay
outputs a number of data values that are significant to the
rendering engine of CryEngine but are largely not useful for
the purposes of THE GRID.
The default navigational controls are: ‘W’ = forward
motion, ‘A’ = leftward motion, ‘S’ = backward motion, and
‘D’ = rightward motion while holding the right mouse button
pivots your view in two angular dimensions (i. e. you can
turn, but not tilt, your head). Near the bottom center of the
viewport is a value that controls speed, next to ‘Speed.’
Control this to move faster or slower. Take some time to get
used to this navigation mode.
Now, while hovering somewhere in the air, press ‘Ctrl
+ G’. You will appear as a guy holding a gun with visible
overlays and fall down into the water. This is the case
because CryEngine is by default an engine for a first person
shooter series of video games and as such has the main
character, controlled by you, hold a gun when he begins
his life. This is one reason to use a template - to set up
CryEngine so that it has you start as a peaceful, gun-less,
overlay-less, civilian takes some steps and it is easier to
start off with them already accomplished.
To navigate, the same WASD + Mouse controls work
here too.
To exit the simulation, press ‘Esc’. Note how the
camera returns to the previous mode at the same spot
where you were as the character. This is called WYSIWYG
(What You See Is What You Get) editing - it preserves real
time and real location of you the controller when shifting
from editing to testing.
9 2
Now, on the main bar up top, under ‘Terrain’ select ‘Edit
Terrain.’ In the new window, above under ‘Modify’ select
‘Set Ocean Height.’ In the pop-up, enter ‘0’ and click ‘OK.’
You can close the terrain editing window.
Now you will see that the ocean has effectively
disappeared. If you press ‘Ctrl + G’ now, you will either fall
to the ground and survive and be able to walk around, or
fall and die. If you die simply press the left mouse button
to reappear where you died. This is close to a template
that would simply have an endless plane of dirt, grass,
concrete, or other simple material.
What is important to note here is that while the ground
you fell on appears flat, it is actually a series of tessellated
triangles. You can see this by pressing ‘F3’ and flying out
until you see some distance across the ground. The area
nearest you will have denser triangles, beyond that, in
large squares, the triangles will merge into larger triangles,
and farthest of all the triangles will be huge and take up
large chunks of space. Pressing ‘F3’ again returns the view
to normal.
9 3
This plane can be molded and shaped into slopes,
hills, mountains, and other geologic features. Integrating
this technology into the design process could be useful
with training and planning (i. e. making room for the terrain
in the original Rhinoceros model). Right now, it will only
come into play as part of various templates. below is an
example of rough slopes quickly painted in CryEngine.
To control the time of day, click on ‘Lighting’ under
‘Terrain’ up top. This will bring up an options window
where you can set the time of day, north offset, latitude,
and tweaks on dawn and dusk times.
So far we have gone over loading a level, navigation,
testing, ocean control, the nature of the terrain, and
controlling sunlight. These are all basic skills and are
enough to effectively use THE GRID to interact in real time
with a design model. Now the only things needed are
loading a template and importing the actual model.
Download the template packages from here. Unzip the
contents of the package file using your favorite zip program
(Explorer can also do it natively) into the \CryEngine\
GameSDK\Levels folder on your machine. Now, either click
on ‘Open’ under ‘File’ or hit ‘Ctrl + O.’ In the window that
pops up, open up the ‘Templates’ folder and choose any
template that you want.
Now enter the simulation (press ‘Ctrl + G’) anywhere
near the ground. You will now see that you are gun-less
and without the overlays.
9 4
The last thing to note is that, when loading a template,
unless you want to tweak the template itself, you should
always save as (in the File menu) when you drop your
models in. If you attempt to save one of the templates, you
will get a warning to remind you.
In the next tutorial we will learn how to import into
CryEngine from 3dsMax.
4. ImportIng
The entire importing process is as follows: a model
in Rhinoceros is exported as an .obj file; that file is
imported into 3dsMax; the CryEngine tools in 3dsMax are
used to export that model into a format that CryEngine can
understand; then the materials are assigned in CryEngine.
The first step is to prepare a project folder that
CryEngine can access. CryEngine uses a separate folder
for each asset, be it an object (a model), a level, or any
other element, many of which are not necessary for THE
GRID. Take a look at the folder contents below:
There are two Rhinoceros files, .3dm and .3dmbak,
one CryEngine object file, .cgf, a 3dsMax file, .max, a
material file, .mtl and an interchange file, .obj. The first four
files (the .3dmbak file is an automatic Rhinoceros backup
file) and the last file contain largely the same information,
namely the 3D information of the model. The remaining file
contains material definitions that 3dsMax exports so that
CryEngine can apply them to the model. Note that the path
is \CryEngine\GameSDK\Objects\[name]. The folder name
must match the name of the .cgf file so that CryEngine,
when browsing for the file, understands what is inside the
folder.
Whenever moving to CryEngine it is useful to either
copy the most updated Rhinoceros file to this folder and
work from it or to export the .obj file from Rhinoceros
directly to the folder so that 3dsMax already has it set as
the project location.
Begin by opening 3dsMax (with Rhinoceros and its
model already open):
4. ImPortIng
9 5
While it is opening, switch to Rhinoceros. One
thing that is important to note is the CryEngine works in
centimeters as the default unit. Whenever you want to
export, switch to centimeters as the model unit using the
command ‘Units’ (accepting the automatic scale). Now,
select what you want to export. Use the command ‘Export’
or find it in the File menu.
Find your project folder and export the file with the
same name as the folder. Use the default settings, as
seen below, making sure that ‘Map Rhino Z to OBJ Y’ is
off. Do not worry about the ‘Export material definitions’
checkbox just below that; material export and handling will
be covered in the next tutorial.
In the next window, again stay with the default setting,
however make sure to later explore the effect of changing
the mesh detail on curved surfaces.
9 6
Now that you have an .obj file return to 3dsMax. From
the default screen, open the menu at the upper left corner
and select ‘Import’.
Find your .obj model and import it with the default
settings. Again make sure ‘Flip ZY-axis’ is off. You will
be presented with your model in 3dsMax. Note that any
NURBS surfaces are now triangulated meshes.
Next, make sure the units are translated properly. One
unit in 3dsMax equals one centimeter in CryEngine, so it
is important that the units in 3dsMax are what they will
be in CryEngine. In the ‘Customize’ menu at top, select
‘Units Setup...’. In the window that appears, set the units
to ‘Metric’ and ‘Centimeters’ and click on the ‘System Unit
Setup’ button. In the new window, set it so that 1 unit
equals 1.0 Centimeters, as below:
Now, press ‘M’. In the new window, first double-click
the only entry in the list on the left, which should be named
something like ‘default (standard) [object_1,object_2,
{etc..}]’. This is a basic material applied to everything you
exported, even if the original objects in Rhinoceros were
from separate layers.
9 7
Next, in the center pane, double-click the object that
has appeared in the blue header. You should now see
something like this:
Next, where it says ‘Blinn’ click and select ‘Crytek
Shader’. In the new options just below, click on ‘Physicalize’
and in the blank drop down next to that select ‘Default’.
Now you can close the material editor. In the next tutorial,
we will explore this window in more detail. For now, what
you just did was tell CryEngine that everything that you
exported is a physical object and must have collision
generated for it. Without this step for everything that is
solid, CryEngine will not create collision and when you test
the model in CryEngine you will simply walk through the
model.
Now, click on the little hammer on the upper right. In
this tab, click on ‘More...’ and in the new window double-
click ‘CryENGINE3 Exporter’. This will open the CryEngine
export settings. To make sure it is properly synced with
CryEngine, click and drag the black scrollbar on the right
down until you reach a closed ‘Options’ section. Click on it
to open it and then click on ‘CryENGINE3 Settings’. Wait a
bit, then in the new window make sure it looks as below:
It may be necessary to hit ‘Scan for builds...’ if it does
not have the CryEngine location set. Close this window and
scroll back up in the pane.
Hit ‘Ctrl + A’ to select everything and click on ‘Add
Selected’ under the ‘Geometry Export’ section. Now scroll a
bit down until you see a blue ‘Export Nodes’.
At this point you must save the 3dsMax model in the
project folder, as otherwise the exporter will not work, if
you have not already saved it. Once saved, click on ‘Export
Nodes’. A window appears and the process should be
9 8
fairly rapid. Once it is finished you can close the window.
Be aware that, once imported to CryEngine, it will
update in real time the model file, so this import process
needs to be setup only once per model, though you still
have to manually export from Rhinoceros and from 3dsMax.
Now, open CryEngine to a template level. In the
RollupBar on the right, click on ‘Brush’. This will show the
browser below. Navigate to your project folder and model.
If nothing went wrong, it should appear. Select it and
drag it in anywhere in the template. It will appear at the
proper scale with a default material that yells ‘Replace Me’.
At this point you can test (Ctrl + G) and walk around
your model. If you want to move the model around, press
the little cross of arrows in the top bar next to the mouse
icon.
In the next tutorial we will see how to replace that ugly
‘Replace Me’ with something else more fitting, as below:
9 9
5. Material Setup
Now it is time to delve into materials and textures.
In Rhinoceros, prepare your materials as usual, except
make sure to name them separate and descriptive things
so that they will be easier to track later on. For now only
set a basic texture for each material; it will be easier to set
other maps in CryEngine anyway. For this tutorial we will
assume you have at least two materials, as the procedure
for more than one material is different than the procedure
for a single material.
Then, export as before, making sure to check the
‘Export material definitions’ option near the bottom of
the export window. This will produce a separate .mtl file
named the same way as your .obj file.
Now, open 3dsMax as before (preparing the units) and
import your model. Make sure ‘Import materials’, ‘Import
into Mat-Editor’, and ‘Show maps in viewport’ are checked
in the lower right section, as below.
The model should import with the materials visible,
3dsMax importing the material file paths as defined in
Rhinoceros. Press ‘M’ as before, when you set up a basic
model. Double-click (or drag into the field) each of your
materials, which should be named as you named them in
Rhinoceros. Physicalize them as before.
The next step is to combine these materials into one
material that CryEngine can understand. What we will
actually do is assign a different material ID to each set
of objects, all of the objects which had the same material
assigned, so that the material within CryEngine can apply
the proper sub-materials to all of the objects. Each sub-
material will be one of the materials in the model. The
model will then have a multi-material applied.
Arrange the materials in a visual way so that you can
easily tell which will be ID 1, ID 2, etc. or simply remember
how they should be, or name them with numbers. By
default all objects have a material ID of 1. So for a model
with more than one material we will have to set all objects
of a material beyond the first with IDs greater than 1
corresponding to which material they have.
For the second material, double-click on the material
and click on ‘Select by Material’ on the end of the top bar,
shown following:
5. materIal setUP
1 0 0
In the new window, each object of that material will
be selected. Hit ‘Select’. The objects with that material will
now be selected. In the right panel, click on the little blue
rainbow and in the ‘Modifier List’ dropdown select ‘Material’
as below:
In the ‘Parameters’ box which appears below, change
the ‘Material ID’ to 2.
Repeat this for each set of objects by material.
Whenever you modify the model (by deleting the imported
geometry in 3dsMax and importing again), you will need
to reassign the material ID but not the material settings in
3dsMax (unless there is something you want to change
about the material). This may get tedious, but there does
not seem to be any way to export the material IDs straight
from Rhinoceros without some kind of script.
Now, right-click in the middle field and select under
‘Materials’ ‘Multi/Sub-Object’. Once it appears, double-click
it and in the options on the right, click on ‘Set Number’. Set
that to how many materials you have. Then, click and drag
in the field from the right little circle of each material to
each little circle on the left of the multi-material in the order
that you determined. Note how the multi-material has ID
numbers set for each material in order and updates when
you connect the materials. An example using two materials
is below:
1 0 1
When you are done, have the multi-material selected
and click on the main viewport. There, select everything
(Ctrl + A). Now, back in the Material Editor, click on the
fourth button from the left in the top bar, ‘Assign Material to
Selection). If the order of the materials with IDs assigned
and the ones connected was the same, the viewport should
not change. If it does, find the error in ID assignment or
connection and rectify it (you can either reassign IDs as
before or change them in the multi-material options).
Now, proceed to export using the CryEngine3 Exporter
utility. Remember to save the 3dsMax session in your
project folder in CryEngine’s folder (\CryEngine\GameSDK\
Objects\).
In CryEngine, open a template level and open the
‘Brush’ menu. Drag your model to the level. By default
it will display ‘Replace Me’ all over it. In the RollupBar
where you just saw the browser, it should have changed
to show various properties of the Brush. Next to where it
says ‘Mtl:’ it should say <No Custom Material>. Click on
this. In the Material Editor that opens up, navigate to your
project folder and note how it has the original exported
.mtl file there. Ignore that and right-click on your project
folder and select ‘Add New Multi Material’. Right-click the
new material and select ‘Set Number of Sub-Materials’. Set
that to the number of materials you have. For each sub-
material that appears under that material, right-click and
select ‘Rename’ and rename them as needed, in order. Your
material editor should appear as below:
Here we can get acquainted with the material editor.
Note how there are several sections: Material Settings,
Opacity Settings, Lighting Settings, Advanced, Texture
Maps, Shader Params, Shader Generation Params, Vertex
Deformation, and Layer Presets. For the purposes of THE
GRID we will only use the first five except for ‘Advanced’.
1 0 2
The first section is only useful as far as setting material
surface properties (how it reacts when you walk on or hit
it). Under ‘Surface Type’ you will find various surface types.
You will set these as needed when you customize each
material.
The second section may be useful for glass and water
with the ‘Opacity’ setting, which is self-explanatory.
Under ‘Lighting Settings’, nearly all settings will be
useful. ‘Diffuse Color’ sets the overall brightness of the
base material texture. ‘Specular Color’ sets the shinyness
color. ‘Glossiness’ sets the spread of shine. ‘Specular
Level’ sets the amount of shine. ‘Emissive Color’ does not
seem to do much. ‘Glow Amount’ is the amount bright parts
of the material will visibly glow.
The last group is where you set the actual textures
directly, by pressing the ‘...’ button.
Now, select the multi material, have your model
selected, and click on the button in the top left, ‘Assign
Item to Selected Objects’. Without texture assignments,
nothing should happen, but now you can play with the
glossiness or opacity settings.
The reason we have not set any textures yet is that
they have to be a certain file type. CryEngine does have a
plugin for Photoshop which allows it to export texture files
specifically set for CryEngine’s use. However, it seems
that Photoshop’s default file saving creates files that work.
The file type that the textures have to be is .tif, which is
available by default in Photoshop’s options for file type
when you save a file.
Open your texture files in Photoshop and save them
as .tif (TIFF, at the bottom) in your project folder. I have
found that a ‘LZW’ compression works; ignore the other
options. Now, for each material, scroll to where it says
‘Diffuse’ under ‘Texture Maps’ and click on ‘...’ to the right of
it. Locate your texture and set it. It may take a few seconds
for CryEngine to import it, but when it does the texture
should appear on your model. If you want, you can set the
other maps, the treatment is similar to how it is in Vray.
Repeat for all of your materials.
If your materials appear too bright, remember that you
can darken the ‘Diffuse Color’ in the third section.
If you want to adjust tiling for each map, you can click
on the black triangle to the left of each map, then click on
the triangle next to ‘Tiling’, and there set both ‘TileU’ and
‘TileV’ to whatever you want.
If you have many materials or many models and you
1 0 3
cannot tell which material is being used where, you can
use the eyedropper button in the Material Editor on the
top bar, fourth from the left, and click on the material in
question in the main view. See below:
One thing to note is that, for some reason, CryEngine
does not update the physical proxy (collision) when you
make changes to the model before exporting from 3dsMax
even though it updates material assignments. So, when
you change your model based on input from THE GRID, re-
export it from 3dsMax, and open it in CryEngine, you will
need to place it again and reassign its material, otherwise
the collision from the previous version will remain. I have
found no other way around this problem.
To recap, the following is needed for the first time
materials are setup:
• Set materials in Rhinoceros. Export to 3dsMax and
drag materials into the Material Editor there.
• Set material IDs for each set of objects based on its
number in order of all the materials.
• Create a multi-material that connects all the materials
in order and assign it to all the objects of the model.
• Export and open in CryEngine and create a multi
material from scratch and assign it, setting all textures.
When changing the model in Rhinoceros, only the
following has to be done:
• Set material IDs for each set of objects based on its
number in order of all the materials.
• Create a multi-material that connects all the materials
in order and assign it to all the objects of the model.
• Export to CryEngine, delete the previous there (even
though it updated in real time) and place the model again,
and reassign the material.
In the next tutorial we will explore how to set up lights.
1 0 4
6. Lighting Setup
In this tutorial we will learn how to work with
CryEngine’s entities and create lights for the imported
model.
There is no way to import lights from Rhinoceros to
CryEngine so the best way is to set up lights directly in
CryEngine.
Open up your model and on
the RollupBar click on ‘Entity’ and
in the browser open ‘Lights’ and
drag a ‘Light’ into your model.
Note that may appear that
your light is invisible. All lights
in CryEngine are by default point
lights (except the sun) and are
merely light emitters, with no
geometry information.
Leave it alone for now, as now is a good time to learn
about helpers. Helpers are visual guides that aid in setting
up CryEngine scenes. Try pressing and holding ‘Shift’ to
show world grids and the selected object’s relation to
those grids. If your light is invisible, hit ‘Shift + Space bar’
to make it appear as a little bulb. Lastly, press and hold the
space bar to show object names.
6. lIghtIng setUP
1 0 5
Now that we have our light and can see it, it is time
to familiarize ourselves with the settings. With the light
selected, examine the ‘Entity Properties’ section in the
RollupBar. Many of the properties are self-explanatory:
Active, Radius, Diffuse (color), DiffuseMultiplier, AreaLight,
PlaneHeight and PlaneWidth.
To setup a realistic light, first change ‘Castshadows’
to ‘LowSpec’. This will ensure that its dynamic shadows
are always available. By adjusting the radius, color
and multiplier, or switching it to an area light, you can
already create a varied set of lighting situations. To
slightly improve the quality of the dynamic shadows, set
‘ShadowResolution’ to 2.
To create a region of shadow or otherwise globally
affect a region without applying shadows from a specific
point, activate ‘Ambient’. The color of the light will now
affect the geometry around it nonspecifically. If you set the
color very near 0,0,0, you can darken areas. This could be
useful to tweak interior spaces that are not directly shaded
by the sun.
If you are making an area light, you can rotate using
the ‘Select and Rotate’ tool in the top bar.
To add a flare to a light, such as bright lights might
create, click on the field next to ‘Flare’ and press the ‘D’
button. This will bring up the Lens Flare Editor. In the upper
left, click on the folder icon and double-click the only entry,
sample_flares.xml. In the left pane, open up the various
categories and select a flare. To apply it to the light you
had selected, click on the fourth button from the left in
the middle portion of the top bar, ‘Assign Item to Selected
Objects’. Close the Lens Flare Editor and click on ‘Flare
Enable’ under ‘Flare’.
‘HDRDynamic’ makes the light brighter in the HDR, so
the engine calculates it relatively to other lights. The sun is
at about 3 for this value.
‘SpecularMultiplier’ makes highlights generated by the
light brighter.
Lastly, to copy a light, or any object, press ‘Ctrl + C’
and move your mouse to where you want the new object
to be. If you press ‘Ctrl + C’ again, CryEngine will drop the
object at the mouse and make a new floating copy that
follows your mouse.
In the next tutorial we will learn about vegetation,
clouds, water, rain, and fog.
1 0 6
7. EnvironmEnt SEtup
CryEngine’s environmental effects cover grass, trees,
clouds, rain and snow, bodies of water outside of the
ocean, global haze and local fog.
Grass and trees are part of the same system, called
Vegetation, and exist in some form in each of the template
levels. For the purposes of THE GRID, we will not go into
detail how to set up grass and trees, however we will learn
how to add more to a level from the ones already existing
in the level.
Begin by opening any template level. In the RollupBar,
select the second tab at the top which looks like a half
pipe. In this tab, click on ‘Vegetation’.
Note how in the lower field there are already entries,
probably called Grass and Trees, or just Grass. Before we
add more, click and drag in the main viewport anywhere
there is grass or trees. This will select those instances.
With this selection you can operate as you would with
an object, you can move it, delete it, rotate it, etc.
Now, to add more, we use a process called painting.
(Incidentally this is similar to how you make terrain). Click
on any of the object entries, click on ‘Paint Objects’ above,
and set the slider under ‘Brush Radius’ to somewhere in
the middle of its bar. Fly up so you can see some ground
and click and drag anywhere in the viewport. Objects of
that type should appear where you paint.
7. envIronment setUP
1 0 7
Note that some objects would not appear under a
certain brush radius. This is because their density is greater
than the set radius, so the probability of them appearing in
the brush area is 0. To delete, hold Ctrl and paint over what
you want gone.
Clouds are also something that is present by default
in the templates and likewise do not need to be explored
in detail.
To add more clouds, switch back to the default tab in
the RollupBar and, under ‘Entity’, navigate to ‘Render’ and
drag one of ‘Cloud’ and ‘VolumeObject’ into your scene.
The Cloud entity is what all of the templates have,
except way up in the sky and scaled to cloud size. The
VolumeObject is supposed to have superior shadowing and
volumetric effects but I have found that it does not really
add much and is generally too bright. To scale either, use
the ‘Select and Scale’ tool on the top bar, where the other
operation tools are.
There are different cloud presets, but the default ones
are fine.
Another way to add clouds is to change the HDR sky
to a skybox, which sacrifices time of day lighting in favor
of a realistic sky scenario. To change to a skybox, click
on the halfpipe tab and then on ‘Environment’. Scroll in
the field below to where it says ‘SkyBox’ and under that
‘Material’ and click on the field on the right, and there press
the ‘..’ button.
This will open the Material Editor. Navigate to
materials>sky and pick anything there. Next, go back to
the RollupBar and in the same field click on ‘<’. This will
apply the selected sky to the scene.
Note that the skyboxes interact only partially with
the sun and will not work well at night. Also, the custom
1 0 8
clouds might not appear to fit. To go back to an HDR sky,
set the material to ‘sky’ in the same material folder.
Rain and snow are handled the same way. Both are
special particle effects, however they are not advanced
enough to interact with geometry (i. e. they still appear
indoors) so it is best to use them only outside.
To add either, we need to open the DataBase View.
Under ‘View’ on the top bar, in Open View Pane, find
‘DataBase View’ and click it.
In this window, click on the ‘Particles’ tab on the top
and then click on the folder icon on the top bar on the left.
Double-click on ‘Libs\particles\weather.xml’. This will load
a folder structure. Open up ‘Rain’ or ‘snow’ and drag either
entry into the level. There are some settings that could be
changed but the defaults are fine.
Rain, above, first appears fairly thick and faint, so it
may be a good idea to scale the droplets down. In the
RollupBar, under ‘ParticleEntity Properties’ find ‘Scale’ and
adjust it as needed.
Snow, above, renders better and may be a good
size visually from the start. You can also try adjusting
‘CountScale’ in the RollupBar, though it may take a few
seconds for changes to appear.
1 0 9
Bodies of water are easy to set up. To create one,
in the default RollupBar pane click on ‘Area’ and then
‘WaterVolume’. Click multiple times in the viewport to draw
a shape for the volume in plan. On the last point, double-
click to finish the shape.
Right now the volume exists but it does not have a
material set. Just like with a model, set the material to
anything in materials>water>watervolumes. The volume
will now become visible. By default it has a thick fog inside
it and a depth of 10. In the ‘WaterVolume Params’ section
in the RollupBar, play with the fog settings and Depth as
you see fit.
To adjust global haze, we need to use the Time of Day
dialog. Find it on the left:
It may be necessary to activate the advanced
properties. It might be useful to rearrange the tabs to
appear as below, to make it cleaner. On the ‘Tasks’ pane,
click on ‘Toggle Advanced Properties’. Then switch back to
‘Parameters’.
Here, scroll down to the Fog settings. The most
important thing to understand here is that any and all of
these values are animated over the length of a day. This is
shown in the graph field on the left. Click around on a few
of the values. Note how many have different curves plotted
over time (top of the graph) from 0 to 24 hours.
1 1 0
Basic operation of the graph window is as follows:
middle-click and hold to move, double-click on the line to
add a point, double-click on a point to delete it, click on a
point to select it, press the red ‘X’ immediately above the
graph with a point selected to clear all points but that one,
and click on the two sine graph icons with brackets in the
top right of the graph to zoom to all the points if you get
lost. You can also click and drag the pink line to adjust
time. It is not as precise as it is in the ‘Lighting’ menu under
‘Terrain’ up top, but can be useful to see the effects over
time.
The most basic settings of the fog that you would
want to change are ‘Global density’ and ‘Height (top)’. They
are self-explanatory, but have a significant impact on the
presence of the haze in the level. Also, ‘Color (bottom)’ and
its multiplier are useful to adjust.
If you want to simulate fuzzy shadows as if the sun
was shining through a fog, scroll all the way to the bottom
and set ‘Shadow jittering’ to some high value.
To create local fog volumes, find the ‘FogVolume’
entity under Entity, Render, the same place where the
clouds were. Drop it somewhere and give it a size. By
default the volume is an ellipsoid; to make it a rectangular
volume, change ‘VolumeType’ to 1. Adjust ‘FallOffScale’ and
‘GlobalDensity’ to change the thickness and softness at
the edges to produce the desired look. I have found that
setting ‘HDRDynamic’ to 1 makes it appear more natural in
1 1 1
response to the sun.
In the next tutorial we will learn terrain editing basics.
8. Editing thE tErrain
In this tutorial we will explore the basics of terrain
painting and layer painting. There are three main areas of
terrain editing: ‘Modify’, ‘Holes’, and ‘Layer Painter’ (and the
separate option of raising or removing the ocean):
These options are under the half pipe tab in the
RollupBar. The first, Modify, is the basic set of brushes
that can rise, lower, smooth, and flatten terrain. Most of
the options are self-explanatory. Play with the settings to
understand the values relative to the unit scale. To switch
from rising to lowering, hold Ctrl. To set a height for ‘Flatten’,
use ‘Pick Height’ at a spot where you want the height to
be. Be careful with using the ‘Enable Noise’ setting: it may
overpower the Rise/Lower switch, making it appear that
you can only raise terrain.
8. edItIng the terraIn
1 1 2
To add an ocean or change the height of the water,
open the Terrain Editor under ‘Terrain’ in the top bar, and
then ‘Edit Terrain’.
In the new window, under ‘Modify’, take note of two
options near the top: ‘Remove Ocean’ and ‘Set Ocean
Height’. These can come in handy if you want to add an
ocean or raise it, or remove the ocean altogether.
The next option is Holes. This tool allows you to cut
out holes in the terrain.
This is useful when your project cuts into a hill or cliff,
or the ground itself. Adjust the brush to cut out larger holes,
and switch to ‘Remove Hole’ when you want to restore the
terrain.
The last option is Layer Painter.
This option allows you to paint different material layers
on the terrain using any custom color and brightness. To
set the material layers themselves, click on ‘Texture’ on the
left, between Terrain and Time of Day.
1 1 3
This will bring up the Terrain Texture Layers window.
In this window, all of the template levels already have
layers set up, but we will go through the steps of loading a
material and texture. Both are necessary because at long
distances the terrain layer uses the layer texture and at
close distances it uses the material, to maximize efficiency
and lessen load on the renderer.
Click on ‘Add layer’ on the left. A new layer will appear
appearing as a grey and white checkerboard pattern. You
can double-click on ‘NewLayer’ to rename it. To set a layer
texture, click on ‘Change Layer Texture’ in the lower left. In
the new window, navigate to GameSDK/textures/terrain (it
should also be a default favorite at the top of the window).
Here pick a default texture. Ideally the layer texture should
be a detailed, almost satellite view, approximation of the
terrain as a whole for that material, but in general these
rough default textures will work.
Now, click on the blue ‘Materials/material_terrain_
default’. In the Material Editor, navigate to materials>terrain
and pick something appropriate. With that selected,
go back to Terrain Texture Layers and click on ‘Assign
Material’. Now the layer is ready to be painted.
In the ‘Layer Painter’ mode, try painting. You will note
that by default the color does not really match what you
want, so play with the color picker and the ‘Brightness’
value until the layer looks like what you want. Examine
it at different times of the day. Also, adjust the ‘Altitude’
and ‘Slope(deg.)’ settings to limit painting only on terrain
patches that meet those conditions. When you settle on a
color for a layer (you can change layers in the lowest field),
click on ‘Save Layer’ to assign that Color and Brightness
combination to that layer by default.
There is actually another way to set color, a way to
bypass having to paint the layer (though it only changes
color, so the material still has to be painted somehow, i.
e. painted first as some color and then this other method
is applied), but it is troublesome and is not necessary for
now.
In the next tutorial we will explore interactive elements.
1 1 4
overvIew of the InteraCtIon tUtorIals
While the first eight tutorials mostly focused on issues
of visualization, the last four focused on interaction and
coaxing presence out of THE GRID. CryEngine offers a
wide range of interactive capabilities, though designing
from an embodied viewpoint was not quite possible. Using
the rudimentary design tools could only be done in the
editor mode, though AI simulation could be run that would
interact with added volumes or volumes that were moved
or stretched.
While within the game mode, agents could be made
to interact with the user if the user moved close to them.
Each time the user enters the game mode the simulation
is different and unique. This presented an interesting
scenario because that meant that not the same model
was experienced from one use to another. Even the same
person on THE GRID on two different occasions would get
a different experience the second time.
Lastly, the interaction extended to environmental
effects, like wind and the sun. The sun could be made to
move at an accelerated pace; this was the setup during the
final review.
1 1 5
9. InteractIve elements
While CryEngine can have a myriad of interactive
actions and events, due to the limitations of THE GRID
they will not be explored here. Instead, we will only look
at some very basic interactions: boids, parting grass, and
water ripples.
Boids are simple agents that look like birds, insects,
and small animals that move around naturally but run away
when you get close to them.
To place some boids, in the RollupBar under Entity
open up the ‘Boids’ folder and place something from there
anywhere in the viewport. That is all that is needed; the
boid will become active by itself.
In its options, there are only two things of great value,
‘Count’ and ‘EnableFlocking’. They are more or less self-
explanatory.
Flocking lets groups of boids generated by one boid
entity to group together as a flock.
Another interactive element is parting grass. This is
not very spatially interactive but is nevertheless a feature
that adds to immersion. The grass models used by most
of the templates do not do this (except for Island, you can
see this effect there), but if you were to paint vegetation
with one of the default CryEngine grass blade models, they
would move away from you when you walked over them
as if you were pushing them aside.
9. InteraCtIve elements
1 1 6
The last interactive element that we will cover is water
ripples. This effect happens automatically when you walk
into either a river (like in the River template) or a body of
water, except for the ocean.
Again, there are many more diverse interactive
elements that can be made, like doors, elevators, wheeled
objects, and so on, but learning those represents an
extensive investment into the system and for the purposes
of THE GRID will not be attempted here.
10. Publishing
In this tutorial we will learn about publishing - packaging
the level so that it can be opened in the Launcher program
(the gateway onto THE GRID) so anyone could explore your
model.
The Launcher is located in Bin32 and Bin64 in the main
CryEngine folder, depending on which fits your operating
system. It is called ‘GameSDK.exe’ and exists separate
from the main editor program.
The first and only thing that needs to happen in the
level that contains your model is that you have to add a
spawn point. This is the location where a user will appear
when loading your level to explore your model.
Locate the ‘SpawnPoint’ under ‘Others’ in the Entity
browser in the RollupBar and drag it in to where you want
a user to begin.
10. PUblIshIng
1 1 7
Note that it looks like a little figure in a box. The box is
very close to six feet in height and a meter square at the
base. The direction the figure is facing is the direction the
user will start in.
The only other step is an option in the file menu. Find
‘Export to Engine’ or hit ‘Ctrl+E’. This will optimize the files
for better loading in the Launcher.
When you load the Launcher, select ‘Levels’ and find
your level. Ignore the templates as they will spawn you
under the ground.
That is it! It is technically possible to strip the entire
CryEngine installation into a portable format, saving only
the files needed for a particular model and level, and then
to setup the Launcher so that it immediately loads the
model, but there is, at the time of writing, no simple way of
doing this. A tool did exist, but it expired.
11. AdvAnced: The designer Tool
The Designer tool is a way within CryEngine to create
simple volumes that are fully colliding and can have
a material applied to them. It may be useful for quickly
blocking out an area.
11. advanCed: the desIgner tool
1 1 8
The Designer tool can be accessed from the RollupBar:
Before we start placing the entity it is important to
make sure the start point will be where we think it will be.
CryEngine needs to be told to recognize when your mouse
is over terrain or a brush, otherwise it will default to the
world grid, which may be far below your terrain. That is
done by a button on the top bar:
There are two buttons: one with a little arrow running
across a ditch and another with the arrow going over a
ball in the ditch. The first is if you want to start on terrain
and the second is if you want to start also on a brush. The
second is usually the better option.
When you click on ‘Designer’ and mouse over the
viewport, you will note that your mouse cursor changes.
From here you can immediately click and drag to create
the base of a shape, ending when you stop pressing your
mouse, and then move your mouse to set the height.
Under ‘Create Brush Parameters’, the tool can be set
to create a small variety of shapes: Box, Cone, Sphere,
Cylinder, Plane, and a custom extruded Shape. For the
Cone, Sphere, and Cylinder there is a setting for the number
of sides.
For example, if you set it to Cylinder and set the Num
Sides to ‘6’:
And drag a shape in the viewport, you will get
1 1 9
something like this:
In the next section, ‘Enter Designer’ is a button labeled
‘Edit Mode’. After clicking this you will be taken to a
geometry editing mode that acts on any Designer shape
you have.
Here are various controls; the really only useful one
you would need is Extrude, which functions similarly to
ExtrudeSrf in Rhinoceros - you select a surface and move
your mouse to extrude. Note that the direction of the
extrusion matters, as extrusions into a shape will make
backward-facing sides which will be invisible from one
side and will not be very realistic. You can experiment with
the other tools to see if one does something useful to you.
The other, more useful, function of the Edit Mode is
that you can select vertices and surfaces in the viewport
and, using the movement and rotation controls on the top
bar, move and rotate them to fine tune a shape.
Pressing ‘Esc’ will take you out of Edit Mode.
There are some tools under the ‘Extra Tools’ section,
but they seem to have problems. In general however, you
will do your design modeling in a separate program.
1 2 0
12. AdvAnced: BAsic Flow GrAph
The Flow Graph is a means within CryEngine of
controlling actions, actors, and events. It is very similar in
style to Grasshopper, if you are familiar with that, or any
other visual coding environment.
The Flow Graph is actually already being
used in each of the templates to remove the HUD
and the player character’s weapon.
To open the Flow Graph, click on the ‘FG’
button on the left. This will open the default Flow
Graph window:
If you are in one of the templates, you can expand
Entities > Defaults > Cloud3 in the file structure on the
left. Clicking on ‘Cloud3’ will load that Flow Graph into the
center pane.
Here you will see four components which are connected:
Game:Start on the left and Inventory:ItemRemoveAll and two
of Debug:ExecuteString. The first component is the event of
the game starting and will execute anything connected to
its output at the start of a simulation. It leads to all three
of the other components. The second component empties
the character’s inventory, removing his gun as he always
has a gun by default. The last two components run two
console commands: hud_hide 1 and godmode 1. The first
command hides the HUD and the second command makes
12. advanCed: basIC flow graPh
1 2 1
the character impervious to damage caused from large falls
(this is to help testing should you fall out of a window or
something).
You can click and drag each component to reposition
them. To connect a component to another, click and drag
from one arrow on one side of a component to an arrow on
the other side of another component. To break a connection,
click and drag the end of a connection (with the arrowhead)
to an empty spot and let go.
To add a component, right-click in the field and select
‘Add Node’. This will produce a huge list of actions, but the
basic operation is simple.
While the Flow Graph is very powerful within
CryEngine, learning its ins and outs would take a very long
time and for the purposes of THE GRID there are only two
uses that are useful to know.
The first use is controlling the rate of the sun and time-
based changes as set in the Time of Day window.
Navigate to ‘Game’ and then click on ‘Start’. That will
add a Game:Start node where you clicked. This is always
needed if you want something to happen at the start of the
game, or to be always true.
Now, add a Time:TimeofDay component. Connect
the output of Game:Start to SetSpeed of Time:TimeofDay.
This means that once the game starts the speed control
1 2 2
of that second component will be activated. That second
component is used to set the time of day and the speed
of the day. To change the speed, either double-click on
‘Speed=1’ or click on the component and on the right
change the number next to Speed. That is it for this Flow
Graph; once you start a simulation the sun will move faster
or slower depending on how you set it. You can also use
this component to set the time of day to a specific value
(just also set the output of Game:Start to SetTime in this
component).
The second use is a basic AI setup. The CFA template
has more advanced AI using the Flow Graph, but explaining
that will be too complex for these tutorials. You are
welcome, however, to examine the Flow Graphs in that
template.
A basic AI setup involves adding a navigation area
where AI will operate, adding a character and a location for
him to move to within that area, adding both as components
in the Flow Graph, emptying his inventory so he’s less
violent (though he will still pretend that he is carrying a gun)
and pacifying him, and telling him to move to that location.
The template Flow Graphs are attached to a cloud in
each template but for this setup we will be more general
and use a FlowgraphEntity, which is just a container for the
AI Flow Graph that we will make. In general, Flow Graphs
need to be attached to something in a level for them to save
with the level.
So, from the RollupBar drag a ‘Human’ entity under AI
> Characters and a ‘FlowgraphEntity’ under Default into the
viewport from the Entity section.
Also, drag a ‘TagPoint’ from the ‘AI’ section. Still in that
section, click on ‘NavigationArea’ and click in the viewport
to draw a shape. Make sure helpers are turned on to see
it (Shift+Space).
Select the NavigationArea and in its RollupBar options
check ‘MediumSizedcharacters’. This is important as it will
tell CryEngine to build a navigation mesh for people-sized
1 2 3
AI (as opposed to vehicles).
Now, click on the FlowgraphEntity and scroll in the
RollupBar until you reach the Flow Graph section. There,
click on ‘Create’.
In the new window, hit ‘New...’ and type in a name
for a group. This name does not matter as long as you
can recognize it. Hit ‘OK’. This will bring up the Flow Graph
window. You will see on the left that there is a new entry
under Entities > [your group name].
Now go back to the viewport. Click on your TagPoint
and in the Flow Graph field right-click and select ‘Add
Selected Entity’. Also right-click and add the AI:GoTo and
Game:Start nodes. In the viewport, select the Human and
in the Flow Graph field right-click on ‘Choose Entity’ in the
first node and select ‘Assign selected entity’.
The only connections you have to make are the
output of Game:Start to Sync of AI:GoTo and the pos of
entity:TagPoint to the pos of AI:GoTo.
That’s it for this setup. To test it within the viewport
you can click on ‘AI/Physics’ to run a simulation of the AI
within the viewport or test it as usual.
The AI will operate as expected if you test in the
viewport, but if you test in game then the character will
react in a hostile manner towards you and try to shoot you.
This is not very useful for THE GRID so we have to
placate him.
The way to do that is to change his faction (factions
are groups of AI that reach in ways to each other) and
empty his inventory (take away his weapon).
Click on the Human and in the RollupBar scroll down
1 2 4
until you reach the ‘Entity Properties’ section.
There are only two thing to change here, EquipmentPack
and Faction. Click twice on the ‘Grunts’ and set it to
‘Civilians’, as below.
Now, click on ‘Player_Default’ right above it and then
on the ‘...’ on its right. In the new window, the only thing
you need to change is the top left dropdown, where you
need to change it to ‘Empty.’
Then click ‘OK’. You will need to do this for every new
Human you add, or just copy them as needed.
That is it for some basic AI. You can try experimenting
with moving the TagPoint and having the characters try to
follow it to get to it, or create spatial challenges for the
characters to attempt to maneuver.
1 2 5
Several basic templates were also developed for work
on THE GRID that were referenced in the tutorials. The idea
was that a user learning THE GRID would try their models
in the templates first before figuring out how to create an
environment from scratch. The templates, of course, could
also themselves serve as usable environments, however a
full project would have a real world context that would need
to be accounted for.
The five templates are: Clean Slate, Island, River,
Road, and Trees and Hills. Each one focuses on one basic
condition that a project may fall into. They use a variety
of CryEngine features and are bare enough to inspire
modification or expansion. These templates start with hints
of experience that can grow with the user’s interaction, and
then direct the user’s imagination.
Clean Slate covers any project that does not, or does
not yet, have a context. It’s basic features are a lack of
features and clouds in the sky. The ground is perfectly flat
for several kilometers, covered only in sparse grass.
Island is a medium sized patch of land in a large
the temPlates
Fig. 5.2 Each of the templates with their main features.
1 2 6
ocean. It is covered in trees that react to wind. The island
itself is a mixture of flat and hilly areas, with the trees
changing density throughout.
River is a modification of Clean Slate because it adds
a moving river, trees, and haze. The wind affects the trees
and the water - objects in the water move with the current.
The haze adds an atmospheric effect and reacts with the
sun.
Road is a simpler modification of Clean Slate. It adds a
long and winding road with two rusty cars by the side. The
road is a tool in CryEngine that can be drawn in any shape
the user wants with any texture, and it sticks to the terrain.
Lastly, Trees and Hills is a hilly plain with trees. Much
like Island, it features a combination of hills and flat areas.
It heavily features trees and their interaction with the sun
and wind.
1 2 7
b e I n g o n t h e g r I d
The fundamental and irrevocable problem with
designing a tool is finding the user for the tool. It is
impossible for the creator of the tool to design the tool to
his or her own metrics and still have the tool register with
other users. It is not the same as it is with designing a
building - in a building subjective decisions are balanced
against a wealth of experience in what works and what
does not. With a tool, one must remain objective to facilitate
the greatest possible audience and user base for the tool
as well as allowing a certain amount of flexibility for the
tool so that it can be diversified and built upon.
With THE GRID, the visualization software will only
be used in the way the users see fit for whatever projects
they use it for. The only way to understand the potential use
is to have users actually use THE GRID and reflect on their
usage. That way the full meta-pipeline of creator-tool-user-
creator-tool will become visible and the even greater thesis
framework will become apparent. For the determination of
the argument depends on the identification of the words
and the structure binding them together. THE GRID is the
words and the structure is the way it used and the way its
dIffICUltIes of the tool
Fig. 6.1 Invitation to get on THE GRID.
1 2 8
use is analyzed.
[][][][][][][][][][]
As the users interacted with THE GRID, I compiled and
documented their interactions. The aim was to determine
if a set of users with varied experience with digital
visualization can gain any benefit from using the tools
of THE GRID and if an impact on the greater architectural
practice can be predicted. If the users find that THE GRID is
only another tool and do not understand how it can supplant
static renders as an experiential prototype, that will tell me
that THE GRID is not yet ready to be used as a design or
as a presentation tool. If they do find that THE GRID adds to
their design process, that will tell me that THE GRID is an
element of the design process that is missing and that can
positively impact the communication that architects and
spatial designers can have with their clients. Of course,
both paths are subjective assessments that I will make as
judgments on my thesis work.
Fig. 6.2 THE GRID is developed only through the addition and accumulation of many separate experiences.
1 2 9
While the users were to go through the design
challenges I planned to perform a parallel series of
developments using THE GRID to revisit my last complete
studio project, the Systems Integration studio, and use THE
GRID to see if a different design emerges because of the
altered viewpoint. That project was a large office building
in the Strip District.
This exploration would have involved me resurrecting
the project files from my archives. I would have used
renders and analysis I had done during the time of the
studio as a control condition. Comparing my use of THE
GRID with this project directly by these means would have
allowed me to see the impact of THE GRID. One caveat is
that THE GRID is inherently real time, so it would have be
en difficult to compare a moving experience to a static one.
However, as the design challenges were developed
this separate exploration was dropped in favor of acquiring
external feedback on THE GRID. Only some Octane renders
were done.
sPrIng 2013 ProjeCt
1 3 0Fig. 6.3 Octane renders of the project. At the time of the studio, Vray and Rhinoceros 4 were not able to render any single image because the geometric detail was too high. Each of these renders was under two minutes.
1 3 1
Each of the users performed a series of short design
challenges using the developed tutorials as a guide as to
how to use THE GRID. Each design challenge used a part
of THE GRID and related it to a style of rendering that was
previously only done via a Rhinoceros -Vray-Photoshop/
Illustrator pipeline. In actuality the users did not follow
the challenges to the letter and used THE GRID in a more
flexible manner. After performing each design challenge,
each user reflected in an online questionnaire on their
experience.
Each of the challenges is reproduced on the facing
page. The challenges were broken down into three key
experiential aspects: the vantage point, temporality, and
activity, respectively titled MOVE, SHIFT, and IMPEL, with
the titles increasingly focusing on a kind of movement that
is implied in THE GRID’s experience. Each challenge was
further broken down into ‘what’, ‘why’ and ‘how’ sections
to facilitate understanding and learning.
The challenges were available as printouts during
testing for the students to peruse.
the desIgn Challenges
Fig. 6.4 Diagrams showing simplified potential results of the challenges. The top shows how a user can place models into a context and navigate around them. The middle shows the same model with searons and time changes. The bottom shows
users interacting with the model.
1 3 2
1 3 3
The testers were provided with a relatively detailed
model of the CFA studio space to perform their challenges
and explorations in. The model was complete with tables,
chairs, computer banks, trash cans, and wood trim. It was
bounded on all sides - the doors were flat, blocked, regions
and the windows were impenetrable.
Since the users accessed the model directly from
CryEngine, they could perform some of the actions the
tutorials detailed, such as controlling the sun and lighting
and placing objects in the space.
Another aspect was the hardware - the School
of Architecture provided a computer station for use of
THE GRID. This made it much easier to get students to
participate because they now no longer had to provide
a computer to use for THE GRID. It also leveled out the
hardware component because everyone used the same
machine. The machine itself, however, was not very good
but THE GRID was still able to run fairly well thanks to the
flexibility of CryEngine.
Following are analyses of the user interactions.
User InteraCtIons
Fig. 6.5 Analyzing the recordings involved retracing the steps of the users, taking into account time and the duration of events, and generally an interplay of factors that are not easily extracted just by looking.
1 3 4Fig. 6.6 A comparison of the hardware the school provided, left, and my own desktop setup, right, both running THE GRID in the studio model. My desktop used the new graphics card that I purchased to use for this thesis, but even overall it was a better set of computer hardware. The major advantage the school computer had was that it was in studio and thus accessible to the students, while my desktop would have had to be lugged over every time it was needed.
1 3 5
The first category of interaction centered around the
user’s cognition. The most important part about seeing on
THE GRID is that it allows for the user to gather data and
gives the user visual cues by which they could respond
to their interaction on THE GRID. This data gathering,
cognition, extends beyond that normally provided by a
static image. In a static image, the general position of the
user is determined through hints: the size of a door, or
the closeness of a wall. On THE GRID, the data gathering
extends towards a displaced proximity, objects that are
nearby but outside the view,solar and environmental
adjustment, and collision in general.
The visual cues that the user sees allow the user
to move and interact. They are the most basic clues
that THE GRID provides - they connect one viewpoint to
another and are needed to make the experience coherent.
The closeness of an object or a distant feature becomes
important when the user is able to move, but reacting
to that cue informs the motion itself. Likewise, objects
disappearing and reappearing give weight to one particular
view over another. As things change, the totality of the
experience is compared to any one position. The user
learns to favor viewpoints.
Fig. 6.7 Examples of user interaction related to responding to visual cues.
1 3 6Fig. 6.8 Frame strip of a continued experience on THE GRID. Here the user navigates through the space, but is sometimes impeded by objects and spatial features.
1 3 7
A deeper level of interaction comes when the user
realizes they are in the space. With this realization comes
a sense of responsibility for the user’s motion and actions.
A sense of the height, the thickness and the mass of the
user sets in. Also, while in motion, the user realizes their
momentum and recognizes that they will take up space in
the area in front of where they are when they move into
it. This becomes extended into a knowledge of the space
where the user falls into learned paths of motion. Motion
becomes emotion. In that way, experience becomes stored
in the model.
Since experience is always personal, this storage
takes a piece of the user and the user leaves that bit
in the model. They become invested - their interaction
becomes precious. This type of caring is both beneficial
and detrimental. Beneficially, it lets the user get used to
the model and find favorite areas, as well as determine
what needs revision based more on the experiences they
have had than the experience they have at that moment.
Detrimentally, they lose sight of the design process and
want to avoid drastic changes to the model.
Fig. 6.9 Examples of user interaction related to attachment to the model.
1 3 8Fig. 6.10 Frame strip of a continued experience on THE GRID. Here the user makes changes to the model which influence the position of the user, producing desirable and undesirable conditions.
1 3 9
To fully allow the experience and the design process
to take full advantage of THE GRID, the creativity of the
user has to be limited. With a CAD viewport, there is too
much to be developed and thought of - the mind does not
know what to focus on, or how to tell what needs creativity
or imagination and what does not. With a static render,
the experience is stifling and constrained - too much has
been taken away from the control of the user, too much is
already defined.
THE GRID strikes a balance between the unresolved
creativity of a CAD viewport and the constraint of a static
render. The focus of the experience is neither on the
geometric and analytical qualities nor on the visual and
photorealism. The experience is partly generated by the
user, partly the result of choices made in the design, but
it is from those choices that the experience develops. By
selecting a narrow aspect of the project’s experience, and
then letting the user control the exposure, creativity is
allowed to grow.
[][][][][][][][][][]
The user experiences, even though limited in hardware,
allowed me to truly see how users responded to THE GRID.
From there, I was able to make my final argument.
Fig. 6.11 Examples of user interaction related to imagination in the model.
1 4 0Fig. 6.12 Frame strip of a continued experience on THE GRID. Here the user experiences limitations within the model, but those limitations allow for imagination, though the amplification of the rest of the model.
1 4 3 Fig. 7.1 The immersion of THE GRID.
Table of Contents 144
Conclusion 145
The Final Review 145
Looking Beyond THE GRID 151
Appendix 155
Sources 155
Terms 164
Part three - table of Contents
t h e f U t U r e o f t h e g r I d
1 4 4
1 4 5
The final review took place in the Miller Gallery in
the Purnell Center for the Arts on the campus of Carnegie
Mellon University. This was the first time, as far as I knew,
that the School of Architecture had hosted the thesis
presentations either in a gallery in general or in a gallery
on campus. The scale and the responsibility represented
in the choice of venue meant that the presentation for THE
GRID could be no simple plot and model setup.
It was agreed between me and my advisors to frame
the presentation as a series of layers. The layers were
both the position of THE GRID in architectural practice and
its position within the study of sight and representation.
They were as follows, from out to in:
VISION – How it looks like. This is the outermost layer
of perception, the initial acquisition of visual data by the
user and their eyes. This is when the user answers “what
it is” without yet understanding any meaning or purpose.
DISCERNMENT – What it means. Once the user
recognizes what it is they are looking at, they can delve
deeper and understand the meaning of what they are
the fInal revIew
C o n C l U s I o n
Fig. 7.2 The initial mock-up of the presentation in the smaller 3rd floor space.
Fig. 7.3 The updated design reflecting the new location.
1 4 6
looking at. This requires a certain amount of involvement
on the part of the user; the designer invests in the design.
AGENCY – Influence and the imagination of the
designer. At this stage the user gains control of the
underlying forces which shape the meaning of the visible.
This is the unpredictable control imparted upon the design
by the designer.
PRESENCE – The designer in the design, his or her
projection of will. This is the thinking and the internal motion
of the designer as it exists and persists in the design. A
user can only reach this level if the design is fully realized
and fully draws the user in.
Those were external layers - each one also
corresponded to an internal layer: in respective sequence:
COGNITION, EMOTION, IMAGINATION, and DASEIN. The
three outermost layers were on plots, while the PRESENCE/
DASEIN layer was THE GRID itself - an interactive station
that anyone could use.
The plots were mounted on boards that I constructed
prior to the presentation. The internal layers were all on
one large board while the external counterparts were on
separate overlapping boards, forcing the user to change
their position to experience them all. They also had
Fig. 7.4 Photos of the development process. The top two on the left show the initial assembly in the CFA building and the Miller Gallery, the top right photo shows two boards assembled in the Miller Gallery, and the bottom photo shows all of the boards assembled but without the plots or bumpouts, or the projection and computer.
1 4 7
bumpouts for each diagram - cardboard extrusions that
enhanced their three-dimensionality.
The presentation days were Friday, April 25th, where
there was an opening reception during which the public and
other students could view all of the projects, and Saturday
and Sunday the 26th and 27th, during which each thesis
student got an hour to present. I went on Sunday.
The presentation changed between Friday and the
weekend. On Friday, there was a monitor set up where
videos played demonstrating the power of THE GRID. On
the wall next to the external plots was a plot summarizing
the point of THE GRID. For the weekend, the monitor went
away and the summary plot replaced it. Due to the layout
of the space, during the opening reception seeing the
monitor as people entered the gallery drew them in, while
over the weekend the core argument of the thesis was
more important so the summary plot became the first thing
to draw people in.
Also, a camera was set up in the corner to record
various interactions of users with THE GRID. It was also
used to record THE GRID breaking down on Sunday
afternoon.
[][][][][][][][][][]
Fig. 7.5 The presentation setup on Friday, during the opening reception.
Fig. 7.6 Details of the presentation setup. The left shows the interactive station, with mouse and keyboard and instructions for use. Behind it is my computer, using the new graphics card I purchased for use with the thesis, with speakers. On the right is a poster summarizing the point of the thesis, which moved places between Friday
and the weekend.
Fig. 7.7 QR code for a short video showing people interacting with THE GRID on Friday.
Fig. 7.8 QR code for a video comparing Vray to THE GRID that was shown on Friday.
1 4 8
The main points from the review were:
•EMBODIMENT - the role of the body in forming ideas.
This approach was brought up as an interesting angle
on my thesis, though due to the technical aspects of my
thesis I never had the chance to explore it, though the
idea of presence and dasein, and immersion in general
approach the concept of embodiment in a roundabout
manner. It was always important during work on my thesis
that the light be shined on the body in the render - that the
designer is not just a disembodied eyeball floating around,
rotating the model or design or building indiscriminately
and omnipotently. It was poignant when, discussing my
thesis with a colleague the lack of the orbit rotation from
Rhinoceros was brought up - on THE GRID, the user moves,
not the model, and this kind of motion, where your position
matters, is natural and real. An orbit mode is unnatural and
even impossible.
•GAMIST - the idea of being gamist was brought up.
This was an odd point that I thought had already been
cleared up, but a new audience at the review brought new
viewpoints. Gamist in this sense means being biased
against video games or their technology without truly
comprehending the possibilities emerging within the video
game engines available today. Efforts like THE GRID, which
Fig. 7.9 Photos of the presentation boards. On top is the left board, which was mental/internal and thus flat. On bottom were the three physical/external boards, which were 3D and changed in space. The two left ones could only be seen if the observer moved.
1 4 9
place video game engines in the context and practical use
of professional practice, should influence gamist attitudes,
teaching by showing that this technology has power and
potential.
•REPRESENTATION AND THE THING - this point pierces
to the core of THE GRID. There is a fine line between the
way something is seen, and the thing itself. What THE GRID
does is it gives meaning to digital models, but at every step
it knows that the digital model is itself a representation,
a representation of a real building not yet built. Therefore
there is a responsibility on the user to realize that they are
not designing the representation using a thing (the digital
model) but using a representation - with user interactions
on THE GRID having an impact, like the ability to knock over
a trashcan or the inability to walk through walls or chairs,
users could always remain present while being distant,
and design the thing directly.
•BEHAVIOR SETTING - the last major point was that,
by using a model of the studio space, I was creating a
model of behavior, as there were people in the audience
who recognized the space and people who did not. There
was a behavioral connection with the first group that made
them feel responsible - and this responsibility thus could be
measured, and algorithmic behavior modeled in response.
Fig. 7.10 Photos of the presentation setup over the weekend. Note how the TV was removed and the summary plot was moved in its place. THE GRID was still fully interactive.
Fig. 7.11 Photo of a sunrise on THE GRID. The lights in the space are turned down, THE GRID takes center stage.
Fig. 7.12 QR code for a time-lapse of the presentation being broken down.
1 5 0
Fig. 7.13 Photos of the rest of the presentation. There were thirteen other students, with projects and ideas ranging the spectrum of architectural practice and theory.
1 5 1
The final review was generally very positive and I was
surprised by how readily people understood THE GRID
once they took the step to inquire. However the review
only presented a reduced version of THE GRID, because
it was only what I could, as a student of architecture at
Carnegie Mellon University, achieve with the means and
funding available to me. The true GRID, the GRID when it
is no longer a GRID, would need to go beyond the realm
of possibility in three categories: the frame, the interface,
and the process.
MAXIMIZATION OF THE DIGITAL FRAME - One of the
fundamental problems and inefficiencies with THE GRID
that I worked with and presented was that it was ultimately
a digital display composed of discrete pixels that could be
counted and amounted roughly to a 1.3 megapixel image.
An average photo has between 4 and 38 megapixels, and
naturally a physical painting or drawing has practically
infinite resolution, though one can argue the resolution
depends on the thickness of the brush fibers or pencil tip.
The magnitude of discrepancy - the average photo is more
than twice as dense resolution-wise - is shown with THE
lookIng beyond the grId
Fig. 7.14 Pixels of THE GRID.
1 5 2
GRID when even a few feet away from the screen, which,
from the interactive station took up about half of the view
of the user, it becomes highly pixelated and breaks its
illusion.
A digital monitor would have the same problem since
the user would constantly attempt to try to get close to
the screen, or otherwise the screen would attempt to get
close to the user - the more of the view THE GRID takes
up, the more immersive it becomes. Pushing the resolution
to a level such that the pixels never become discernable
would be a challenge: the eye has about 576 megapixels
of resolution with a small focus point where the eye sees
about 7 megapixels. Very high resolutions are possible
with CryEngine, as it is possible to tell the engine to render
at over 26 megapixels. However, with more pixels comes
more load on processing power, and the more needs to be
rendered the stronger a graphics card is required for a real
time view.
EMERGING INTERFACES - A key direction THE GRID did
not take was advanced interfaces - either motion tracking
or tactile joysticks, both with a virtual reality headset. THE
GRID as it was presented relied on a keyboard and mouse,
both because they were the only interfaces available and
because the typical design firm would be sure to have Fig. 7.15 Section of a 7680 x 4320 output resolution shot from CryEngine.
1 5 3
them. The true GRID, however, would involve the user
being unbound from a computer station and would connect
the motion of their body with the motion of their user on
THE GRID. That means a motion of an arm in real life would
move their arm on THE GRID. A pulling motion would create
form on THE GRID, a pointing motion would draw curves,
and so on.
While the body is being directly connected to the
design control, the eyes are presented with a fully
immersive 3D display within a virtual reality headset. This
headset would have a high-resolution image over each
eye, simulating binocular vision. If the system is haptic
enough, the user would eliminate the technological middle
man and experience the design with true embodiment - the
body and the eyes would see the design directly as if it
was real life.
Such technology, like the very high resolition, is
currently possible but not yet reachable. Headsets are in
development by Oculus Rift and Sony that can seamlessly
integrate with other interfaces, as long as the renderer can
output two images a few inches apart. It is awareness
of technology just like this that THE GRID as a thesis
attempted to generate among rising architects.
RENDERING PRECEDES DESIGN - The last and far
Fig. 7.16 Snapshot of a video Untold Games released of their in-development game Loading Human, modeled on Unreal Engine 4. This game is designed from the ground up to be used with a VR headset and motion joysticks. While the technology still limits the user to a seated position, it is not a great leap of thought to imagine it tracking fully body motion in a large open space, such as an aircraft hangar or an open field.
1 5 4
reaching aspect of THE GRID was truly applying it to
design. What happens when an architect and a client
use THE GRID from beginning to end, becoming used to
a photorealistic representation (is it representation at that
point, if it is all they see?) and then experiencing the thing
in real life? Does it become disappointing, because it is no
longer perfect, or because it ages, or because habitation
and cognitive offloading give meaning to it that could not be
transferred on THE GRID?
This would represent a fundamental shift in design
process - the rendering would now precede design, as
knowing what it looks like would be more important than
knowing what it is. That means the design would always
have to be seen before it is thought of, a process that is
even more creative than that of dream, where the sight and
the thought are simultaneous. With the need for an image,
or imagery in general, to exist first, the creative force would
accelerate unimaginably. Combined with the previous
categories, design would almost literally spring from the
fingertips of the designer, fully and imprecicely resolved
before it actually exists.
The real would cease to exist. Time will only tell what
will remain...or will time itself be gone? The future is bright,
and the dawn comes. See you beyond THE GRID.
Fig. 7.17 THE GRID would completely upset architect-client-contractor dynamics. Each sector would have their own idea, their own developed experience, of the reality represented by THE GRID.
1 5 5
BOOKS AND RESEARCH REPORTS
Darley, Andrew. Visual Digital Culture: Surface Play and Spectacle in New Media Genres. London ; New York: Routledge, 2000.
Dieter Hildebrandt, Jan Klimke, Benjamin Hagedorn, and Jürgen Döllner. 2011. Service-oriented interactive 3D visualization of
massive 3D city models on thin clients. In Proceedings of the 2nd International Conference on Computing for Geospatial Research
& Applications (COM.Geo ‘11). ACM, New York, NY, USA, , Article 6 , 1 pages. DOI=10.1145/1999320.1999326 http://doi.acm.
org/10.1145/1999320.1999326
Emiliyan Petkov. 2010. One approach for creation of images and video for a multiview autostereoscopic 3D display. In Proceedings
of the 11th International Conference on Computer Systems and Technologies and Workshop for PhD Students in Computing on
International Conference on Computer Systems and Technologies (CompSysTech ‘10), Boris Rachev and Angel Smrikarov (Eds.). ACM,
New York, NY, USA, 317-322. DOI=10.1145/1839379.1839435 http://doi.acm.org/10.1145/1839379.1839435
Heiko Friedrich, Johannes Günther, Andreas Dietrich, Michael Scherbaum, Hans-Peter Seidel, and Philipp Slusallek. 2006.
Exploring the use of ray tracing for future games. In Proceedings of the 2006 ACM SIGGRAPH symposium on Videogames (Sandbox
‘06). ACM, New York, NY, USA, 41-50. DOI=10.1145/1183316.1183323 http://doi.acm.org/10.1145/1183316.1183323
Jongeun Cha, Mohamad Eid, and Abdulmotaleb El Saddik. 2009. Touchable 3D video system. ACM Trans. Multimedia Comput.
Commun. Appl. 5, 4, Article 29 (November 2009), 25 pages. DOI=10.1145/1596990.1596993 http://doi.acm.org/10.1145/1596990.1596993
Lewis, Rick. Generating Three-dimensional Building Models From Two-dimensional Architectural Plans. Berkeley, Calif.: University
of California, Berkeley, Computer Science Division, 1996.
so u R C e s
a P P e n D I x
1 5 6
Luc Leblanc, Jocelyn Houle, and Pierre Poulin. 2011. Component-based modeling of complete buildings. In Proceedings of Graphics
Interface 2011 (GI ‘11). Canadian Human-Computer Communications Society, School of Computer Science, University of Waterloo,
Waterloo, Ontario, Canada, 87-94.
Mitrovic, Branko. Visuality for Architects: Architectural Creativity and Modern Theories of Perception and Imagination. University
of Virginia Press, 2013.
Rhyne, Theresa-Marie. “Computer Games and Scientific Visualization.” Association for Computing Machinery. Communications
of the ACM 45.7 (2002): 40-4. ProQuest. Web. 24 Sep. 2013.
Robina E. Hetherington and John P. Scott. 2004. Adding a fourth dimension to three dimensional virtual spaces. In Proceedings of
the ninth international conference on 3D Web technology (Web3D ‘04). ACM, New York, NY, USA, 163-172. DOI=10.1145/985040.985064
http://doi.acm.org/10.1145/985040.985064
YOUTUBE AND VIMEO
alvaroignc. (2010, March 17). Zumthor’s Thermae of Stone in Source SDK part 5: Props. [Video File]. Retrieved from http://www.
youtube.com/watch?v=hh4nGEAKm4s - Zumthor’s Therme Vals rendered in Source.
Archimmersion. (2010, June 25). UDK - Family House in Realtime 3D [Video File]. Retrieved from http://www.youtube.com/
watch?v=AV802r_Pr0k&feature=youtu.be - More UDK - again, note the cheap quality.
Autodesk. (2011, April 12). Autodesk Showcase 2012 for Architectural, Construction, and Engineering Users - YouTube [Video File].
Retrieved from http://www.youtube.com/watch?v=ioP0CVRJvUI#t=17 - This is for reference - this is a very bad implementation of the subject of my thesis as it provides no presence, no true interactivity and is not at all designed for the user.
bigkif. (2007, November 17). Ivan Sutherland : Sketchpad Demo (1/2) [Video file]. Retrieved from http://www.youtube.com/
watch?v=USyoT_Ha_bA
bigkif. (2007, November 17). Ivan Sutherland : Sketchpad Demo (2/2) [Video file]. Retrieved from http://www.youtube.com/
1 5 7
watch?v=BKM3CmRqK2o - Ivan Sutherland’s 1963 Sketchpad thesis, archival footage.
EliteGamer. (2012, November 28). Luminous Engine - Live Edit Tech Demo “Agni’s Philosophy” [Video file]. Retrieved from http://
www.youtube.com/watch?v=eHSGBh1z474 -Luminous Engine tech demo.
GameNewsOfficial. (2013, March 29). Metal Gear Solid 5 Fox Engine Tech Demo [Video file]. Retrieved from http://www.youtube.
com/watch?v=_18nXt_WMF4 -Fox Engine tech demo.
gametrailers. (2012, June 7). Unreal Engine 4 - GT.TV Exclusive Development Walkthrough [Video file]. Retrieved from http://www.
youtube.com/watch?v=MOvfn1p92_8 -Unreal 4 tech demo.
Hammack, David. [hammack710]. (2013 January 3). Unity 3D Simulation Project [Video File]. Retrieved from https://www.youtube.
com/watch?v=EEA5_he3pRk - A demo of Unity3D , looks very cheap and old.
HD, RajmanGaming. (2013, August 21). CryEngine Next Gen (PS4/Xbox One) Tech Demo [1080p] TRUE-HD QUALITY [Video file].
Retrieved from http://www.youtube.com/watch?v=4qGK5lUyCwI -CryEngine demo reel.
Inc, Marketing Department Ideate. (2013, February 26). Autodesk Showcase 3D Visualization Software [Video file]. Retrieved from
http://www.youtube.com/watch?v=IvBL2kX6CME -Autodesk Showcase video.
lxiguis. (2012, August 28). Real time Architectural Visualization - After Image Studios [Video File]. Retrieved from http://www.
youtube.com/watch?v=HPtQyBDpatg&feature=youtu.be - UDK demonstration. It is not that great and a little old, but is a capable engine.
Lapere, Samuel. [SuperGastrocnemius]. (2012, April 6). Real-time photorealistic GPU path tracing: Streets of Asia [Video File].
Retrieved from http://www.youtube.com/watch?v=gZlCWLbwC-0
Lapere, Samuel. [SuperGastrocnemius]. (2013, August 13). Real-time path tracing: 4968 dancing dudes on Stanford bunny [Video
File]. Retrieved from http://www.youtube.com/watch?v=huvbQuQnlq8
1 5 8
Lapere, Samuel. [SuperGastrocnemius]. (2012, May 29). Real-time photorealistic GPU path tracing at 720p: street scene [Video
File]. Retrieved from http://www.youtube.com/watch?v=evfXAUm8D6k -GPU path trace method demonstrations. This is a highly realistic rendering method, short of the grainy appearance.
Lumion3D. (2010, November 1). Architectural visualization: Lumion 3D software is easy to use [Video file]. Retrieved from http://
www.youtube.com/watch?v=uoLV8QIm02M -Demonstration of Lumion 3D.
Naing, Yan. [MegaMedia9]. (2013, May 31). Realtime 3D Architectural Visualization With Game Engines [Video file]. Retrieved from
http://www.youtube.com/watch?v=uXzy3V3N2uw -CryEngine3 demonstration in a sandbox environment.
Skiz076. (2012, January 3). FallingWater in Realtime 3d (UDK) [Video File]. Retrieved from http://www.youtube.com/
watch?v=QdF4rvw64rg - A model of Fallingwater in UDK.
spacexchannel. (2013, September 5). The Future of Design [Video File]. Retrieved from http://www.youtube.com/watch?v=xNqs_S-
zEBY#t=134 - Video showcasing tactile hardware interaction. This is the future, but we are not then yet.
Storus, Matt. (2011, February 9). Video Game Engine Architectural Visualization Test [Video File]. Retrieved from http://vimeo.
com/19774547 -Another CryEngine3 demonstration.
T.V., Arocena. [arocenaTM]. (2011, February 17). Presenting Architecture through Video Game Engine [Video File]. Retrieved from
http://www.youtube.com/watch?v=S8HUj85Cq1s - Demo by Max Arocena with CryEngine showing interactive lighting.
Timeshroom. (2013, July 30). Architectural Visualisation - Oculus Rift Demo [Video file]. Retrieved from http://www.youtube.com/
watch?v=gaFZH8Z70vk -Oculus RIFT demo showing the views provided by the headset. Note how they are slightly offset, this would produce the illusion of 3D.
Visual, Real. [RealVisual3D]. (2012, October 23). iPad 4th Generation: Unity 3d Realtime Architectural Visualisation [Video file].
Retrieved from http://www.youtube.com/watch?v=n6eb4KB2k2U -iPad demonstration of Unity3D and how it is cross platform.
1 5 9
ARTICLES
(2014, May 8). Answering the Unanswerable: What is the Resolution of the Human Eye?. Peta Pixel. Retrieved from http://
petapixel.com/2014/03/12/answering-unanswerable-whats-resolution-human-eye/ -Discussion on the resolution of a human eye.
(2013, August 20). Arch Virtual releases architectural visualization application built with Unity3D game engine, including Oculus
Rift compatibility. Arch Virtual. Retrieved from http://archvirtual.com/2013/08/20/arch-virtual-releases-architectural-visualization-
application-built-with-unity3d-game-engine-including-oculus-rift-compatibility/ -Arch Virtual’s interactive app.
(2013, August 20). Arch Virtual. Retrieved from http://www.archvirtual.com/Panoptic/2013-08-19-arch-virtual-panoptic.html - Premade realtime visualization demo by Arch Virtual. It is interactive within a web browser. This is a very good example of the subject of my thesis.
(2013, June 3). Arch Virtual. Retrieved from http://archvirtual.com/2013/06/03/tutorial-ebook-now-available-unity3d-and-
architectural-visualization-1-week-preview-edition-discount/ - Arch Virtual’s ebooklet on architectural visualization in Unity3D.
(2014, May 8). Crysis 8K resolution hack offers a peek at the next decade of gaming. Engadget. Retrieved from http://www.
engadget.com/2014/05/06/crysis-8k-resolution-hack/ -Article about the CryEngine modification that allows it to render large resolution images.
Elkins, James. (2010, November 6). How Long Does it Take To Look at a Painting? Huffpost Arts & Culture. Retrieved from http://
www.huffingtonpost.com/james-elkins/how-long-does-it-take-to-_b_779946.html - Article showing how Mona Lisa visitors spend 15 seconds looking at it.
(2014, May 8). How many megapixels do you need?. Connect. Retrieved from http://connect.dpreview.com/post/1313669123/
how-many-megapixels -Discussion of the photo resolutions of various phones.
Hudson-Smith, Andrew. digital urban. Retrieved September 2, 2013, from http://www.digitalurban.org/ (deprecated page: http://
www.digitalurban.blogspot.com/) - Blogging platform that publishes research about connecting digital modeling and the real world with an emphasis on the profession of architecture.
(2014, May 8). Hyperrealistic virtual reality adventure Loading Human headed to Oculus Rift and Project Morpheus.
Engadget. Retrieved from http://www.engadget.com/2014/05/07/loading-human-rift-morpheus/?utm_campaign=socialflow&utm_
1 6 0
source=fb&utm_medium=fb -Article that mentions the game developed specifically for a VR headset, with footage.
Jobson, Christopher. (2013 September 22). Full Turn: 3D Light Sculptures Created from Rotating Flat Screen Monitors at High
Speed. Colossal. Retrieved from http://www.thisiscolossal.com/2013/09/full-turn-light-sculpture/?src=footer - A project using alternate projection - this is useful because hardware exploration is part of my thesis, though here the technology is very artsy.
Kasperg. “Kaufmann House.” The Whole Half-Life. 1/23/2006, Retrieved September 2, 2013, from http://twhl.info/vault.
php?map=3657 -Website of the Fallingwater digital recreation. This establishes a kind of benchmark for the possibilities of the area.
Putt, K. Crysis 3 - Alternative. 2014. https://secure.flickr.com/photos/k_putt/13974809675/in/set-72157644191959442/. -Flickr image of CryEngine outputting to a very high resolution.
Russo, Luigi. Architectural Visualization. Unreal Engine. Retrieved September 3, 2013, from http://www.unrealengine.com/
showcase/visualization/architectural_visualization_1/ - Website of a project done in UDK. This is in place to be licensed (educational use included).
simulation. (n.d.) Random House Kernerman Webster’s College Dictionary. (2010). Retrieved October 20, 2013, from http://www.
thefreedictionary.com/simulation - Definition of simulation.
Varney, Allen. “London in Oblivion.” The Escapist. 7/8/2007, Retrieved September 2, 2013, from http://www.escapistmagazine.
com/articles/view/issues/issue_109/1331-London-in-Oblivion - Article that mentions several attempts to visualize architectural work in video game engines. This could be a good springboard on collating past efforts in this area.
Vella, Matt. (2007, December 21). Unreal Architecture. Bloomberg Businessweek. Retrieved from http://www.businessweek.com/
stories/2007-12-21/unreal-architecturebusinessweek-business-news-stock-market-and-financial-advice - Article detailing the use of UDK for architectural purposes.
Wikipedia contributors, “Architectural Animation,” Wikipedia, The Free Encyclopedia, http://en.wikipedia.org/wiki/Architectural_
animation (accessed November 29, 2013). - Wikipedia article on architectural animation.
(2014, May 8). What is the resolution of the human eye?. Science Blogs. Retrieved from http://scienceblogs.com/
cognitivedaily/2006/10/22/what-is-the-resolution-of-the/ -Another discussion on the resolution of a human eye.
1 6 1
IMAGES
Act-3D. 19 April 2012. Lumion logo. [logo]. Retrieved from http://lumion3d.com/forum/general-discussion/lumion-logo/?action=dl
attach;attach=8515 - Lumion logo.
alexglass. 11 October 2013. Ray Tracing vs Rasterized. [chart]. Retrieved from http://www.ign.com/boards/threads/generation-8-
starts-with-brigade-not-x1-ps4.453427233/ - Chart of raster vs. ray tracing technologies.
Blender Foundation, The. n. d. Blender logo. [logo]. Retrieved from http://download.blender.org/institute/logos/blender-plain.
png - Blender logo.
Chaos group. n. d. Vray logo. [logo]. Retrieved from http://upload.wikimedia.org/wikipedia/fa/a/a1/Vray_logo.png - Vray logo.
CryEngine. n. d. CryEngine logo. [logo]. Retrieved from http://www.n3rdabl3.co.uk/wp-content/uploads/2013/08/logo_vertical_
black.jpg -CryEngine logo.
Epic Games. n. d. UDK logo. [logo]. Retrieved from http://epicgames.com/files/technologies/udk-logo.png - UDK logo.
Euclideon. 22 November 2011. Euclideon Unlimited Detail. [screenshot]. Retrieved from http://media1.gameinformer.com/
imagefeed/featured/gameinformer/infdetail/infpower610.jpg - Euclideon screenshot.
Fatahalian, Kayvon. n. d. Kayvon Fatahalian. [graph]. Retrieved from http://www.cs.cmu.edu/~kayvonf/ - Photo of Kayvon Fatahalian.
Fatahalian, Kayvon, et al. July 2013. Visualization graph. [graph]. Retrieved from http://graphics.cs.cmu.edu/projects/
exhaustivecloth/ - Kayvon’s exhaustive graph.
History Blog, The. n. d. Dome design. [drawing]. Retrieved from http://www.thehistoryblog.com/wp-content/uploads/2013/01/
Dome-design.jpg - Brunelleschi’s dome image.
1 6 2
IGX Pro.com. n.d. Mario 64. [screenshot]. Retrieved from http://www.igxpro.com/wp-content/uploads/2012/09/mario64.jpg -Mario64, an old 3D video game.
Jean-Philippe Grimaldi, et al. n. d. LuxRender logo. [logo]. Retrieved from http://upload.wikimedia.org/wikipedia/commons/f/f5/
Luxrender_logo_128px.png - LuxRender logo.
Konami. 27 March 2013. Title. [logo]. Retrieved from http://babysoftmurderhands.com/wp-content/uploads/2013/04/FOX-Engine-
Kojima-Productions-GDC-2.jpg - Comparison of the Fox Engine to real life.
Mh. 10 March 2010. The Gates-Hillman Complex. [photo]. Retrieved from http://upload.wikimedia.org/wikipedia/commons/a/a6/
CMU_Gates_Hillman_Complex.jpg - Photo of the Gates-Hillman Center.
n. d. Tom Cortina. [photo]. Retrieved from http://sigcse2014.sigcse.org/authors/ - Photo of Thomas Cortina.
Otoy, Inc. 22 November 2012. Octane Render logo. [logo]. Retrieved from http://en.wikipedia.org/wiki/File:Octane_Render_logo.
png - Octane logo.
PcGamesHardware. n.d. Crysis 2 screenshot 5. [screenshot]. Retrieved from http://www.pcgameshardware.com/screenshots/
original/2010/03/crysis-2-screenshots-gdc-2010__5_.jpg -Crysis 2 image.
Persage. 5 April 2007. Carnegie Mellon University College of Fine Arts building. [photo]. Retrieved from http://upload.wikimedia.
org/wikipedia/commons/3/3a/CFA.JPG - Photo of the College of Fine Arts.
Unity Technologies. n. d. Unity logo. [logo]. Retrieved from http://upload.wikimedia.org/wikipedia/ru/a/a3/Unity_Logo.png - Unity logo.
MISCELLANEOUS
Adobe & Touch. n. d. Projects Mighty & Napoleon. Retrieved from http://xd.adobe.com/mighty/notify.html - Website of Adobe Mighty and Napoleon.
Autodesk. n. d. 3D visualization software brings design to life. Retrieved from http://www.autodesk.com/products/showcase/
1 6 3
overview - Website of Autodesk showcase.
Crydev. (2013, October 18). CRYENGINE® Free SDK (3.5.4) [Computer software]. Retrieved from http://www.crydev.net/dm_eds/
download_detail.php?id=4 - CryEngine3 SDK.
Lumion. (2013). Lumion 3D () [Computer software]. Retrieved from http://lumion3d.com/ - Lumion website, note how a new version is available, but the evaluation version of it is not yet.
NHTV University of Applied Sciences. (2013, November 11). ARAUNA2 demo. [Computer software]. Retrieved from http://ompf2.
com/viewtopic.php?f=5&t=1887#p4233 - Arauna2 demo.
Schroeder, Scott A. (2011, January 1). Adopting Game Technology for Architectural Visualization. Purdue e-Pubs. Retrieved from
http://docs.lib.purdue.edu/cgi/viewcontent.cgi?article=1005&context=cgttheses - Possible precedent thesis.
1 6 4
3D - A digital representation of three point perspective that approaches how the eyes interpret light. 3D is often mapped onto a
planar screen, but newer technologies are using curved screens, or even a screen for each eye to even more closely replicate vision.
ACM - Association for Computing Machinery.
Animation - In the context of this thesis, refers to a disembodied flythrough that architects are so fond of - an unnatural
movement that lacks artistic merit and generally does not approach human experience. Animations can and do exist that give the
viewer an experience they can understand, but technology can go beyond that.
AO - Ambient occlusion. A technique that replicates GI shading by determining where deep corners are and shading them
accordingly. Combined with other effects, this is an efficient method to fake radiosity shading.
Architecture - The study of the memory of time and space. Encompasses the thought, theory, tools, design, construction,
evaluation, and history of buildings.
Baking - Taking pre-computed data and turning it into a texture that can be applied in a material.
BIM - Building Information Modeling. A type of modeling not necessarily visual that digitally covers architectural systems.
Bump map - A bump map is either another name for a normal map or refers to a greyscale image that appears like the grain or
small-scale detail of a material that is applied to a material in the scene to very efficiently fake said detail. A bump map is the simplest
way to add complexity to a mesh on a small scale by only using a material.
CAD - Computer Aided Design. Digital precision tools used in product, aviation, automotive, and architectural design.
Compute capability - A ranking of CUDA technology, roughly the version number, that relates to how well the CUDA cores can
Te R m s
1 6 5
process their tasks.
CUDA - A technology Nvidia developed for their graphics processors that uses parallel processing that developers can directly
access for graphics purposes.
CMU - Carnegie Mellon University. This is my university and where the School of Architecture is - where I am having my thesis.
Delay load - a term I came up with that describes the relative time it would take to use one program or pipeline compared to
another. For the purposes of the software evaluations, I compared a regular pipeline of modeling in Rhinoceros and rendering in Vray
to each set of alternative software.
DIY - Do it Yourself. A field of development not necessarily informed by professional practice where users attempt to find their
own ways to achieve a task. These attempts are not always successful but the culture is one of sharing - the attempts that work are
often documented and refined.
Drivers - Software middle-men between hardware on a computer and other software that aims to use that hardware.
Engine - A graphics software (that can also be embedded in other software) that is used to render virtual worlds. In video games,
this is what makes the graphics work, though it is often also responsible for physics calculation, the menus and UI, and AI.
Environment map - A single snapshot of the six cardinal directions around an object with a FOV of 90° that are then composited
to get a 360° view completely around an object. This is used to fake reflections. Doing this in real time is very taxing on performance.
FOV - Field of view. The geometric angle that is subsumed by the view cone of a viewer.
Fps - Frames per second, also frame rate. A measure of the amount of frames a graphics processor can generate on a monitor
every second to simulate fluid motion. Values between 30 and 60 are good goals for graphics-heavy software, as at lower values
choppiness and stuttering become apparent and higher values may produce incompatibility with the monitor hardware (usually not
an issue with modern software). This can be measured as an average over the last few seconds or as a value every few seconds.
1 6 6
Gameplay - The actions a user performs in relation to the environment or other players within a game. People often fail to make
the distinction between graphics and gameplay, as one or the other may define a video game more than the other. For this thesis, I am
ignoring almost all aspects of gameplay except those involving interaction, walking, and other movement controls.
GI - Global Illumination. This refers to an even distribution of light in a scene such that more exposed surfaces get more light and
less exposed surfaces get less light. This ends up making corners darker and smoothly shading other geometry. This is useful as a step
in generating realistic shadows.
GPU - Graphics Processing Unit. The piece of hardware in a computer largely responsible for computing what is seen on a monitor.
Over the years the GPU has grown in importance, not only for video games but for design number crunching as well.
GWAP - Games With A Purpose. Video games designed or heavily repurposed for the aim of training real jobs. These video games
are high fidelity and take into account nearly all aspects of a real world scenario. They often focus less on graphics, however.
Mesh - A set of connected or related triangles in 3D space that combine to make a virtual shape or surface. The triangles are solid,
however their appearance can change when an image, or a texture, is applied to the mesh via predefined operations, a material, using
coordinates assigned to each point of the triangles. Meshes can have billions of triangles.
Normals and normal map - A normal is the perpendicular direction from a plane; in meshes the planes are the triangles. A normal
map is a purple and green image that replicates height data, which is projected along the normals of the mesh. This fake height
data appears as ridges or other shapes, depending on the map, that receive lighting and shading but are only a visual effect on the
geometry - it is clipped by the visible edges. A technique called parallax mapping or displacement mapping works around the clipping,
appearing to make physical geometry on top of the original mesh.
NURBS - Non Uniform Rational Basis Spline. A mathematical method for defining a curve that can also be used to define complex
surfaces. Since the definition is mathematical, the surfaces are exact, though a given graphics program approximates the surface with
a mesh for preview purposes. The mesh simply takes a small number of points on the surface and connects them, but the mesh is no
longer the NURBS surface, it is just a very near approximation. Many methods exist to sample those points.
1 6 7
Path tracing - A method of ray tracing that determines where the photons that comprise a pixel most likely came from, taking
into account all the light in a scene. Over enough samples, path tracing should generate an image indistinguishable from reality.
Photo-realistic - A digital simulation that is visually very near or indistinguishable from a photo taken by a real camera.
Pre-computed - Computed before hand, usually a process that takes many minutes or hours, but the results are reusable.
Raster - Very general term for taking a mathematically perfect form and simplifying it for viewing. Raster can refer to pre-
computing shadows in a scene and baking them into the materials in the scene instead of having the shadows be dynamic.
Ray tracing - A graphical technique where photons from lights are traced around a scene, taking into account all possible material
properties, to determine how that scene is lit.
Real time - A digital refresh, or frame, rate at which point the screen looks fluid, like a movie. Reality is in real time.
Render - A technique where a graphical algorithm is applied to a scene that generates how that scene would look, usually, in real
life. It is also a general term for creating a high quality image, so many realistic paintings could be understood as renders.
Rhinoceros - NURBS modeling software developed by McNeel and Associates that is primarily used for nautical, product, and
architectural design. It is fairly streamlined and includes hundreds of functions. Supports scripting and plug-ins.
Scene - A set of geometry, lights, materials, effects, and other features that combine to be used for rendering or interaction.
Design software either imports files to combine into a scene, or saves the scene as a file which references other files.
Shader - A rapid computational process where visual effects like refraction, bumpy sufraces, and reflection are processed as
materials that can be applied to geometry. Shaders are much cheaper than brute force methods but rely on environment maps and
fairly complex material definitions to replicate how these effects appear in real life. Depending on the software, they allow behavior
that would otherwise be difficult to replicate, for example a material can fade depending on how close the viewer is to it.
Shadow map - Precomputed shadows that are applied to all geometry. Shadow maps are stored as color image files, depending
1 6 8
on the lights in the scene, that are then used (usually automatically) in the material shaders of the scene geometry; this is called
baking. Just like with other textures, they use object UV coordinates.
SIGGRAPH - Special Interest Group on Graphics. An annual conference held by ACM that reviews and publishes research on
computer graphics.
Simplified lighting - The use of simple models of how light propagates in space. This ranges from a linear hotspot/falloff model,
with 100% light in a small sphere of an arbitrary radius and 0% light in a larger sphere, and a linear gradient in between, to more complex
models where certain shapes are achieved on surfaces that mimic how real lenses distribute light.
Ultimatype - Direct opposite of prototype - what the object or space will eventually be.
User interaction - The concept of a person using controls on a device to change how that device operates, often this feedback is
displayed on a monitor or screen.
Vector - A mathematically defined curve. Vector graphics have infinite resolution, but cannot exist in real life, so they have to be
turned into a raster image. Likewise, digital photons are also vectors, but they have to be turned into bright spots and dark spots on
surfaces for a user to understand them.
Vertex lighting - An alternate method of generating shadows in a scene. Vertex lighting applies a color value to each vertex of a
geometry that corresponds to the color of the shadow or the light at that spot. Geometry is sometimes subdivided for this purpose to
have a more even distribution of points. The advantage this has over regular shadow maps is that it is not pixel based and will always
have smooth shadows, but at the potential cost of detail.
Vray - Rendering suite developed by Asgvis. Features a fast dual-renderer pipeline that incorporates material definitions, lights, a
sun and sky, caustics, and has support for crude animation. Exists as a plugin for Rhinoceros and other modeling programs.
WYSIWYG - What You See Is What You Get. A design concept where the visual development of something is exactly what that
thing would look like once it is finished. Microsoft Word is a good example of a WYSIWYG program.
This page intentionally left blank.