Siggraph’s technical papers are at the core of what makes SIGGRAPH such a great conference. Sure there are the parties, the cool tech etc., but it is the technical papers that continue to deliver the core of CG development. We asked a few leading industry experts (Weta, SPI, Pixar etc) to tell us what they are personally looking forward to seeing.
Siggraph technical papers
Pixar have long been a huge supporter of SIGGRAPH and Tony DeRose of Pixar Animation Studios is this year’s Technical Papers Chair. “SIGGRAPH 2010 will feature a vibrant field of technical presentations, we are most excited by the extraordinary breadth of topics as well as the fascinating achievements in many fields from architecture to photography.”
DeRose along with the Siggraph papers team have once again put together a great program of key papers. For every paper delivered in LA this year, there is a huge amount of work done both by researchers writing and submitting work, but also by the review teams that judge the large number of submissions received each year. The Technical Papers program is a premier international conference for disseminating new scholarly work in computer graphics. To be accepted, submitted papers must be “novel (they cannot overlap substantially with any paper previously accepted for publication or conference), and they must adhere to the highest scientific standards,” states the Siggraph bylaws and guidelines. “We are looking for high-quality research papers that introduce new ideas to the field and stimulate future trends. In addition to the core topics of modeling, animation, rendering, imaging, and human-computer interaction, we encourage submissions from areas related to computer graphics including: computer games, scientific visualization, information visualization, computer-aided design, computer vision, audio, and robotics…excellence of the ideas is the predominant acceptance criterion.”
Along with our own humble opinions we asked some leading R&D experts and industry heavyweights for their personal opinions and put together this suggestion of must see papers at this year’s SIGGRAPH. This list is not definitive and if we could recommend one thing it would be to make the time to see as many technical papers as possible. These papers really are the intellectual backbone of the conference and the strength of that backbone carries the body of the conference each year.
Of course it is extremely difficult to decide what to see, or even what to recommend – the industry has an embarrassment of riches at this conference, as Dana Batali, head of the RenderMan software team comments, “I’m interested in over 50% of the papers. This always happens to me at SIGGRAPH! Too much to absorb, not enough time! I plan to read through most of these papers and if time allows, to attend the talks. Then after SIGGRAPH, reread the ‘hottest’ papers again.” Batali, as a developer of rendering software, tends to focus on rendering-related tracks however points out, “I’ve also found that venturing into apparently unrelated presentations can be quite fruitful. I tend to place new algorithms & techniques ahead of GPU-optimization and keep hoping that this will be the year when NPR (non-photorealistic rendering) for animation comes of age.” He explains, “In over 20 years of attending SIGGRAPH conferences, I’ve found that keeping an open mind is absolutely crucial. Sometimes the best ideas are hidden in obscure corners – and this include talks, panels, posters and courses!”
To help you we asked a few friends of fxguide at Pixar, ICT, Weta, Mental Images and elsewhere to highlight some of the papers they are personally interested in seeing.
(Of course all of the comments below are just personal opinions and by no means exhaustive!)
Paper’s Fast Foward
One of the most helpful events is the Papers overview presented at the Sunday Fast Forward. Each year at SIGGRAPH, the paper authors provide brief overviews of their work at the Technical Papers Fast Forward event. Later in the week each speaker will present their complete papers in 22.5-minute sessions that include 4.5 minutes of Q&A, but the Sunday Fast Forward is a brilliant chance to get a taste of what is on offer and sets you up really well for the week to not miss any key papers. Nothing is worse than hearing after some session, how great that paper was you missed last night!
Technical Papers Fast Forward
SUNDAY, 25 JULY | 6 – 8 pm
Technical Papers are published as a special issue of ACM Transactions on Graphics. In addition to papers selected by the SIGGRAPH 2010 Technical Papers Jury, the conference presents papers that have been published in ACM Transactions on Graphics during the past year.
Paul Debevec, USC’s Institute for Creative Technologies
Paul Debevec leads the graphics laboratory at the University of Southern California’s Institute for Creative Technologies. Debevec has pioneered high dynamic range image-based lighting techniques in his films Rendering with Natural Light (1998), Fiat Lux (1999), and The Parthenon (2004); he also led the design of HDR Shop, the first high dynamic range image editing program.
At USC ICT, Debevec has developed a series of Light Stage devices for recording the appearance and reflectance properties of human faces used in creating photoreal digital actors in movies such as Spider Man 2 (2004), Superman Returns (2006), and The Curious Case of Benjamin Button (2008) as well as 2008′s “Digital Emily” project.
The Frankencamera: an Experimental Platform for Computational Photography
Stanford has been having fun with cameras again, and plans to invite others to join in with their “Frankencamera” platform for computational photography.
Paper Link
NETRA: Interactive Display for Estimating Refractive Errors and Focal Range
Need glasses from working on the computer? This mobile camera app (with a simple computational photography lens) will tell you your prescription.
Unstructured Video-Based Rendering: Interactive Exploration of Casually Captured Videos
Like Photosynth, evolved to apply to video collections of the same event. Computer vision continues to accomplish new feats.
Sebastian Sylwan, Weta Digital
Sylwan is Chief Technology Officer at Weta Digital. In this role, he is helping to continue Weta Digital’s long tradition of Innovation in VFX through artist aware research and development.
Before joining Weta, Sylwan served as Autodesk’s Senior Film Industry Manager. In this role he helped set the strategy for Autodesk’s products in the global film market, including the first steps into stereoscopic 3D. Previously, he was Director of Technology at Digital Domain. He successfully helped DD venture into new areas like the development of the facility’s first stereoscopic rendition of a CG animated film. Sylwan has also served as Principal Technology Advisor at USC’s Institute for Creative Technologies where he led the Light Stage 6 project.
Example-Based Wrinkle Synthesis for Clothing Animation
Huamin Wang, Florian Hecht, Ravi Ramamoorthi, James O’Brien (University of California at Berkeley)
There are a few example based/data driven real-time cloth papers. This is an interesting direction to go. If I had to pick one, I would probably choose this one.
Reducing Shading on GPUs using Quad-Fragment Merging
Paper Link
Kayvon Fatahalian’s paper is interesting because it demonstrates another step towards micropolygon rendering on GPUs.
High-quality single-shot capture of facial geometry
Paper Link
This one caught my eye, straight from the title. The results look very promising, and even though some of the refinement steps may not necessarily be readily adaptable to production, it definitely sets a good path. I can see potential here.
Physics-Inspired Topology Changes for Thin Fluid Features
url(http://www.cc.gatech.edu/~wojtan/thin_fluid_features/thin_fluid_features.html, Paper Link)
I really like the approach that this paper takes, using physical observation as a basis, and preserving small scale details with consistent topology.
Temporal Upsampling of Performance Geometry using Photometric Alignment
Paper link
And of course ( I may be partial) but Paul’s work is always inspiring, and points in very interesting directions.
Since I’m a rendering nerd (my job) but also have some side-interest in physics simulations such as fluid dynamics, collision, etc.
Zap Andersson, Mental Images
Zap Andersson, Shader writer and product expert at Mental Images on, of course, Mental Ray.
Zap is a very well-known industry expert with numerous talks and speaking engagements to his credit, he also maintains his own blog on all things shader related.
If one starts by simply going through the “trailer” Siggraph video, the things I found most interesting were the Vector Solid Textures, the Hair Shading bits, the volume scattering and brushed metal, but also the audio generation for brittle fractured surfaces (I used to be into audio production back in the day) and the smoke simulation bit.
Actual papers I will really try to catch:
A Coaxial Optical Scanner for Synchronous Acquisition of 3d Geometry and Surface Reflectance
Real world measured data really intrigue me, and is very useful.
A Radiative Transfer Framework for Rendering Materials With Anisotropic Structure
This is right up my alley of interest.
A System for Directoinal Occlusion for Fast Cinematic Lighting of Massive Scenes
Again, this is totally my kind of thing. Looks to be really interesting stuff!
Accurate Multidimensional Poisson-Disk Sampling
Sampling is one of the most fundamental parts of CG, and advances in it is always interesting.
Acquisition and Analysis of Bispectral Bidirectional Reflectance and Reradiation Distribution Functions
- Going beyond RGB and into the venue of spectra has always been interesting. This goes further and into the avenue of frequency shifting light (such as in optical whiteners in cloth and UV fluorescence) – cool stuff!
An Artist Friendly Hair Shading System and Interactive Hair Rendering Under Environment Lighting
Hair is one of those things that is tricky to do right because our brains are so fine-tuned to seeing it day in and day out.
By-Example Synthesis of Architectural Textures
Looks cool.
Effects of Global Illumination Approximations on Material Appearance
Is a perceptually based paper and I am very interested in perception and how to trick the eye to perceive something as “real.”
Fabricating Spatially-Varying Subsurface Scattering, Physical Reproduction of Materials With Specified Subsurface Scattering as well as Line-Space Gathering for Single Scattering in Large Scenes
All look like something to check out for future SSS shaders!
Manifold Bootstrapping for SVBRDF Capture
This sounds very interesting as it applies to real-world materials spatial behaviour.
Modeling and Rendering of Impossible Figures
Back to perception and optical tricks again – I love it!
Photorealistic Models for Pupil-Light Reflex and Iridial Pattern Deformation
Having just recently written an “Eye Shader” as an example of MetaSL programming, this paper is quite interesting to me.
Being a sound nerd, the Rigid-Body Fracture Sound with Precomputed Soundbanks sounds (pun intended) quite … shattering (pun also intended). There are also a couple of other sound related papers I will check out if I have time, since it really doesn’t apply to my work, but to my nerdy interests… 
Toward Evaluation Material Design Interface Paradigms for Novice Users – UI is always a tricky thing so this looks worthwhile to check out!
That’s what sounds most interesting from the titles. As always, at SIGGRAPH, getting time to actually view them all is quite tricky, since they often overlap or collide with some other function/meeting/whatnot one needs to do, which is sad but unavoidable, I guess.
Rob Bredow, Sony Pictures Imageworks
Rob Bredow is Chief Technology Officer and Visual Effects Supervisor at Sony Pictures Imageworks.
As CTO, Bredow has brought a production perspective to the role and also created Imagework’s open source initiative – most recently releasing Open Shading Language to the community. Prior to the CTO position, Bredow supervised Cloudy with a Chance of Meatballs. His other film, Surf’s Up, was nominated for an Academy Award® and was also recognized with two Annie Awards for its groundbreaking visual effects.
While at Imageworks, Bredow has been involved in creating many of the complex visual effects featured in The Polar Express, Stuart Little 2, Cast Away, and Academy Award® nominated Stuart Little . Bredow’s other credits include Godzilla, Independence Day and several others.
Here are the papers that caught Bredow’s eye:
A Coaxial Optical Scanner for Synchronous Acquisition of 3D Geometry and Surface Reflectance
I’m very interested in the state of the art in measuring BRDF’s these days.
Ambient Point Clouds for View Interpolation
I’m curious if this might have application for hole filling and touch ups as it relates to re-dialing stereo cameras in post (adjusting interocular).
An Artist-Friendly Hair Shading System
Making an easy to use hair shader is always tough–and how can you pass up a hair shading paper where Jensen is one of the contributers.
Collision-Free Construction of Animated Feathers Using Implicit Constraint Surfaces
Feathers are really tricky (we experienced this first-hand years ago on Stuart Little 2).
Filament-Based Smoke With Vortex Shedding and Variational Reconnection
Sounds like it might make some great looking images. Speaking of images, I hope they bring a lot of pictures because I probably won’t be able to understand the math.
PantaRay: Fast Ray-Traced Occlusion Caching of Massive Scenes
I’m very interested in their GPU-based solution and how it scaled for Avatar.
The last of these papers also caught our attention, John Montgomery in particular.
fxguide
What caught the eye of fxguide’s John Montgomery:
PantaRay: Fast Ray-Traced Occlusion Caching of Massive Scenes
Jacopo Pantaleoni and Timo Aila, Nvida research. Luca Fascione and Martin Hill, Weta Digital.
I am looking forward to this paper on precomputation of sparse directional occlusion caches used in the context of a fast cinematic lighting pipeline. The system was used as the primary lighting technology during the making of the movie Avatar by the guys at Weta. It was able to handle the massive scenes of unprecedented complexity in that film. The paper seems to be covering GPU and rapid cinematic lighting cycles at final feature film quality by means of a custom spherical harmonics occlusion precomputation pass. A spherical harmonic is a particularly efficient way to store directional information. (It is actually very close, conceptually, to a fourier transform). This approach has shown that it is capable of handling something as complex as Avatar. Looks impressive, and I have taken a real interest in GPU pipelines lately.
What caught the eye of fxguide’s Mike Seymour:
Importance Sampling for Production Rendering
Mark Colbert, Simon Premoze, Guillaume François
I will bend the rules and pick a technical course, as I have been doing heaps of work lately on HDRs/ BRDFs and Filtered Importance Sampling and there is a cracker of a Siggraph Course on this (- but covering some new innovative methods – so it is almost a technical paper!) This course looks like it will not only be a great primer on the state of the art, but it will tie in several new technologies – I’m really looking forward to it.