Upload
cameraculture-mit-media-lab
View
8.626
Download
1
Embed Size (px)
Citation preview
Ramesh Raskar, MIT Media Lab
After X, what is neXt
Coming up with New Ideas in Imaging
Ramesh Raskar, MIT Media Lab
Ramesh Raskar, MIT Media Lab
Xd
X++
X X+Y
X
X
neXt
Ramesh Raskar, MIT Media Lab
Raskar, Camera Culture, MIT Media Lab
Camera Culture
Ramesh Raskar
Camera CultureMIT Media Lab http://raskar.info
http://cameraculture.info
Ramesh RaskarAssociate Professor
Create tools to better capture and share visual information
The goal is to create an entirely new class of imaging platforms
that have an understanding of the world that far exceeds human ability
and produce meaningful abstractions that are well within human comprehensibility
Ramesh Raskar, MIT Media Lab
Camera CultureCamera Culture
Course WebPage :
http://cameraculture.info/courses/
Ramesh Raskar, MIT Media Lab
After X, what is neXt
Coming up with New Ideas in Imaging
Ramesh Raskar, MIT Media Lab
Ramesh Raskar, MIT Media Lab
Xd
X++
X X+Y
X
X
neXt
Ramesh Raskar, MIT Media Lab
Ramesh Raskar, MIT Media Lab
Simple Exercise .. Simple Exercise ..
What is neXt
Ramesh Raskar, MIT Media Lab
Strategy #1: XStrategy #1: Xdd
• Extend it to next dimension (or some other) dimensionExtend it to next dimension (or some other) dimension
• Context aware resizing Context aware resizing – VideoVideo– Instead of square resizing-> CD cover (with a hole in center) resizingInstead of square resizing-> CD cover (with a hole in center) resizing
• Text, Audio (Speech), Image, Video .. Whats next ?Text, Audio (Speech), Image, Video .. Whats next ?
• Video, 3D meshes, 4D lightfieldsVideo, 3D meshes, 4D lightfields• Images to infrared, sound, ultrasoundImages to infrared, sound, ultrasound• Macro scale to microscale (Levoy, Lightfield to Microscope)Macro scale to microscale (Levoy, Lightfield to Microscope)• Time to space to angle to idTime to space to angle to id• (coded exposure <- coded aperture)(coded exposure <- coded aperture)
Coded-Aperture ImagingCoded-Aperture Imaging
• Lens-free imaging!Lens-free imaging!• Pinhole-camera Pinhole-camera
sharpness,sharpness,without massive light without massive light loss.loss.
• No ray bending (OK for No ray bending (OK for X-ray, gamma ray, etc.)X-ray, gamma ray, etc.)
• Two elementsTwo elements– Code Mask: binary Code Mask: binary
(opaque/transparent)(opaque/transparent)– Sensor gridSensor grid
• Mask autocorrelation is Mask autocorrelation is delta function (impulse)delta function (impulse)
• Similar to MotionSensor ?Similar to MotionSensor ?
Flutter Shutter CameraFlutter Shutter CameraRaskar, Agrawal, Tumblin [Siggraph2006]
LCD opacity switched in coded sequence
Figure 2 results
Input Image
Problem: Motion Deblurring
Ramesh Raskar, Camera Culture, MIT Media Lab
Image Deblurred by solving a linear system. No post-processing
Blurred Taxi
Ramesh Raskar, Camera Culture, MIT Media Lab
Flutter Shutter: Shutter is OPEN and CLOSED
Preserves High Spatial Frequencies
Sharp Photo Blurred PhotoPSF == Broadband Function
Fourier Transform
Coded Aperture CameraCoded Aperture Camera
The aperture of a 100 mm lens is modified
Rest of the camera is unmodifiedInsert a coded mask with chosen binary pattern
Out of Focus Photo: Coded Aperture
Captured Blurred Photo
Refocused on Person
Larval Trematode WormLarval Trematode Worm
Ramesh Raskar, MIT Media Lab
Strategy #2: X+YStrategy #2: X+Y• Fusion of the dissimilarFusion of the dissimilar
– More dissimilar, more spectacular the outputMore dissimilar, more spectacular the output
• ExampleExample– Scientific imaging + PhotographyScientific imaging + Photography
• Coded apertureCoded aperture• TomographyTomography
• Lightfields + User interfacesLightfields + User interfaces• Projector = cameraProjector = camera
– Spatial Augmented RealitySpatial Augmented Reality
Ramesh Raskar, MIT Media Lab
Imaging in Sciences: Imaging in Sciences: Computer TomographyComputer Tomography
• http://info.med.yale.edu/intmed/cardio/imaging/techniques/http://info.med.yale.edu/intmed/cardio/imaging/techniques/ct_imaging/ct_imaging/
Ramesh Raskar, MIT Media Lab
Borehole tomographyBorehole tomography
• receivers measure end-to-end travel timereceivers measure end-to-end travel time• reconstruct to find velocities in intervening cellsreconstruct to find velocities in intervening cells• must use limited-angle reconstruction method (like must use limited-angle reconstruction method (like
ART)ART)
(from Reynolds)
Ramesh Raskar, MIT Media Lab
Prototype cameraPrototype camera
4000 4000 × 4000 pixels ÷ 292 × 292 lenses = 14 × 14 pixels per lens× 4000 pixels ÷ 292 × 292 lenses = 14 × 14 pixels per lens
Contax medium format camera Kodak 16-megapixel sensor
Adaptive Optics microlens array 125μ square-sided microlenses
Ramesh Raskar, MIT Media Lab
Ramesh Raskar, MIT Media Lab
Example of digital refocusingExample of digital refocusing
Coded-Aperture ImagingCoded-Aperture Imaging
• Lens-free imaging!Lens-free imaging!• Pinhole-camera Pinhole-camera
sharpness,sharpness,without massive light without massive light loss.loss.
• No ray bending (OK for No ray bending (OK for X-ray, gamma ray, etc.)X-ray, gamma ray, etc.)
• Two elementsTwo elements– Code Mask: binary Code Mask: binary
(opaque/transparent)(opaque/transparent)– Sensor gridSensor grid
• Mask autocorrelation is Mask autocorrelation is delta function (impulse)delta function (impulse)
• Similar to MotionSensor ?Similar to MotionSensor ?
Mask in a Camera
Mask
Aperture
Canon EF 100 mm 1:1.28 Lens, Canon SLR Rebel XT camera
Ramesh Raskar, MIT Media Lab
Strategy #3: X Strategy #3: X Do exactly the oppositeDo exactly the opposite
• Processing, Memory, BandwidthProcessing, Memory, Bandwidth– In Computing world, in any era, one of this is a bottleneckIn Computing world, in any era, one of this is a bottleneck– But overtime, they change. You can often take an older idea and do But overtime, they change. You can often take an older idea and do
exactly the opposite.exactly the opposite.– E.g. bandwidth is now considered virtually limitlessE.g. bandwidth is now considered virtually limitless
• In imaging:In imaging:– Larger sensors?Larger sensors?
• Everyone is thinking about building cheaper, smaller pixel sensors and THEN Everyone is thinking about building cheaper, smaller pixel sensors and THEN improving SNR .. Maybe just build larger sensors?improving SNR .. Maybe just build larger sensors?
– SLR: Faster mirror flip or no mirror flipSLR: Faster mirror flip or no mirror flip• Companies spent years improving mirror flip speedCompanies spent years improving mirror flip speed• Why not just remove it?Why not just remove it?
• More computationMore computation• Less lightLess light
Ramesh Raskar, MIT Media Lab
• e.g. Reverse Auctione.g. Reverse Auction
Less is MoreLess is MoreBlocking Light == More InformationBlocking Light == More Information
Coding in Time Coding in Time Coding in SpaceCoding in Space
Larval Trematode WormLarval Trematode Worm
Mitsubishi Electric Research Laboratories Special Effects in the Real World Raskar 2006
Vicon Motion Capture
High-speed IR Camera
Medical Rehabilitation Athlete Analysis
Performance Capture Biomechanical Analysis
Towards ‘on-set’ motion capture
• 500 Hz with Id for each Marker Tag• Visually imperceptible tags + Natural lighting• Unlimited Number of Tags• Base station and tags only a few 10’s $
Traditional: High-speed IR Camera + Body markers
Second Skin: High-speed LED emitters+ Photosensing Body markers
Mitsubishi Electric Research Laboratories Special Effects in the Real World Raskar 2006
R Raskar, H Nii, B de Decker, Y Hashimoto, J Summet, D Moore, Y Zhao, J Westhues, P Dietz, M Inami, S Nayar, J
Barnwell, M Noland, P Bekaert, V Branzoi, E Bruns
Siggraph 2007
Prakash: Lighting-Aware Motion Capture UsingPhotosensing Markers and Multiplexed Illuminators
Mitsubishi Electric Research Laboratories Special Effects in the Real World Raskar 2006
Imperceptible Tags under clothing, tracked under ambient light
Hidden Marker Tags
Outdoors
Unique Id
Mitsubishi Electric Research Laboratories Special Effects in the Real World Raskar 2006Labeling Space(Indoor GPS)
Each location receives a
unique temporal code
But 60Hz video projector
is too slow
Projector
Tags
Pos=0
Pos=255
Time
Mitsubishi Electric Research Laboratories Special Effects in the Real World Raskar 2006
Pattern
MSB
Pattern
MSB-1
Pattern
LSB
For each taga. From light sequence, decode x and y
coordinateb. Transmit back to RF reader (Id, x, y)
0 1 1 0 0 X=12
Mitsubishi Electric Research Laboratories Special Effects in the Real World Raskar 2006
Inside of Multi-LED Emitter
Mitsubishi Electric Research Laboratories Special Effects in the Real World Raskar 2006
Tag
Ramesh Raskar, MIT Media Lab
• When life gives you lemon, make lemonadeWhen life gives you lemon, make lemonade
Ramesh Raskar, Karhan Tan, Rogerio Feris, Ramesh Raskar, Karhan Tan, Rogerio Feris, Jingyi Yu, Matthew TurkJingyi Yu, Matthew Turk
Mitsubishi Electric Research Labs (MERL), Cambridge, MAMitsubishi Electric Research Labs (MERL), Cambridge, MAU of California at Santa BarbaraU of California at Santa Barbara
U of North Carolina at Chapel HillU of North Carolina at Chapel Hill
Non-photorealistic Camera: Non-photorealistic Camera: Depth Edge Detection Depth Edge Detection andand Stylized Rendering Stylized Rendering
usingusing Multi-Flash ImagingMulti-Flash Imaging
Depth Discontinuities
Internal and externalShape boundaries, Occluding contour, Silhouettes
Depth Edges
Our MethodCanny
Canny Intensity Edge Detection
Our Method
Photo Result
Car Manuals
What are the problems with ‘real’ photo in conveying information ?
Why do we hire artists to draw what can be photographed ?
Shadows
Clutter
Many Colors
Highlight Shape Edges
Mark moving parts
Basic colors
Shadows
Clutter
Many Colors
Highlight Edges
Mark moving parts
Basic colors
A New Problem
Ramesh Raskar, MIT Media Lab
Strategy #4: X Strategy #4: X • Given a Hammer ..Given a Hammer ..
– Find all the nailsFind all the nails– Sometimes even screws and boltsSometimes even screws and bolts
• Given a cool solution/technique, Given a cool solution/technique, – find other problemsfind other problems
• Good recent examplesGood recent examples– Gradient domain techniquesGradient domain techniques
• Introduced in Graphics for High dynamic range tone Introduced in Graphics for High dynamic range tone mapping [Fattal Lischinski 2002]mapping [Fattal Lischinski 2002]
• Now a major hammerNow a major hammer– Image editing, compositing, fusion, alpha matting, reflection layer recoveryImage editing, compositing, fusion, alpha matting, reflection layer recovery
A Night Time Scene: Objects are Difficult to Understand due to Lack of Context
Dark Bldgs
Reflections on bldgs
Unknown shapes
Enhanced Context :All features from night scene are preserved, but background in clear
‘Well-lit’ Bldgs
Reflections in bldgs windows
Tree, Street shapes
Background is captured from day-time scene using the same fixed camera
Night Image
Day Image
Result: Enhanced Image
Flash Result Reflection LayerAmbient
Flash and Ambient ImagesFlash and Ambient Images[ Agrawal, Raskar, Nayar, Li Siggraph05 ][ Agrawal, Raskar, Nayar, Li Siggraph05 ]
Agrawala et al, Digital Photomontage, Siggraph 2004
Agrawala et al, Digital Photomontage, Siggraph 2004
actualphotomontageset of originals
perceived
Source images Brush strokes Computed labeling
Composite
Ramesh Raskar, MIT Media Lab
Strategy #5: X Strategy #5: X • Given a problem, find other solutionsGiven a problem, find other solutions
– Given a nail, find all hammersGiven a nail, find all hammers– Sometimes even screwdrivers and pliers may workSometimes even screwdrivers and pliers may work
• High Dynamic Range Tone MappingHigh Dynamic Range Tone Mapping– Started with Jack Tumblin’s LCISStarted with Jack Tumblin’s LCIS– Gradient domainGradient domain– Bilateral filterBilateral filter– Filter banks etc .. Filter banks etc .. – About 6 years of heavy machineryAbout 6 years of heavy machinery– Btw, the topic is done to death but continues to enthuseBtw, the topic is done to death but continues to enthuse
Ramesh Raskar, MIT Media Lab
Strategy #6: X++Strategy #6: X++• Pick your adjective ..Pick your adjective ..• Making it faster, better, cheaperMaking it faster, better, cheaper
neXt = adjective + XneXt = adjective + X
Ramesh Raskar, MIT Media Lab
X++ : Add your favorite adjectiveX++ : Add your favorite adjective• Context aware, Context aware, • AdaptiveAdaptive• (temporally) Coherent, (temporally) Coherent, • Hierarchical, Hierarchical, • ProgressiveProgressive• EfficientEfficient• ParallelizedParallelized• DistributedDistributed
• Good example: Image or video compression schemesGood example: Image or video compression schemes• But X++ is a bad signBut X++ is a bad sign
– The field is dying in terms of research but booming in business impactThe field is dying in terms of research but booming in business impact
Ramesh Raskar, MIT Media Lab
PitfallsPitfalls• These six ways are only a start These six ways are only a start • They are a good mental exercise and will They are a good mental exercise and will
allow you to train as a researcherallow you to train as a researcher• Great for class projectsGreat for class projects• But But
– Maynot produce radically new ideasMaynot produce radically new ideas– Sometimes a danger of being labeled incrementalSometimes a danger of being labeled incremental– Could be into ‘public domain ideas’Could be into ‘public domain ideas’
Ramesh Raskar, MIT Media Lab
What are Bad ideas to pursueWhat are Bad ideas to pursue• X then Y (then Z)X then Y (then Z)
– X+Y is great with true fusion, fusion of dissimilar is bestX+Y is great with true fusion, fusion of dissimilar is best– But avoid a ‘pipeline’ systems paper, where the output of But avoid a ‘pipeline’ systems paper, where the output of
one is THEN channeled into the input of the next stage, one is THEN channeled into the input of the next stage, and non of the components are noveland non of the components are novel
– E.g. I want to build a E.g. I want to build a • Follow the hype (too much competition) Follow the hype (too much competition) • Do because it can be done Do because it can be done
– (Why do we climb? because it is there! (Why do we climb? because it is there! – But only the first one gets a credit. But only the first one gets a credit. – May make you strong, and give you a sense of May make you strong, and give you a sense of
achievement but not a research project. )achievement but not a research project. )
Ramesh Raskar, MIT Media Lab
Xd
X++
X X+Y
X
X
neXt
Ramesh Raskar, MIT Media Lab
Raskar, Camera Culture, MIT Media Lab
Camera Culture
Ramesh Raskar
Camera CultureMIT Media Lab http://raskar.info
http://cameraculture.info