Seeds of Excitement
Jun 1, 2005 12:00 PM, By Audrey Doyle
SIGGRAPH 2005 Papers
The shadows in this ray-traced image were computed using a new physically-based algorithm that will be discussed in a paper entitled "Soft Shadow Volumes for Ray Tracing."
(Courtesy of Samuli Laine, Helsinki University of Technology)
Since the first SIGGRAPH conference 32 years ago, its continuing mission has been to be the premier annual conference on the leading-edge theory and practice of computer graphics and interactive techniques. And one of the ways the organization has fulfilled that mission every year since 1973 is through the Papers presentations. This year SIGGRAPH received 461 submissions for the Papers program, of which the committee accepted 98 Papers for publication.
Heightened reality and advanced physics simulation are among the hot subjects being covered in the SIGGRAPH 2005 Papers sessions. To get more detail on how these an other topics are being covered in this year's Papers Program, we recently sat down with this year's SIGGRAPH 2005 Papers Chair, Markus Gross, to gather some of his thoughts. Gross is a professor of computer science and director of the Computer Graphics Laboratory of the Swiss Federal Institute of Technology (ETH) in Zürich.
A technique for simulating highly-turbulent fluid effects such as smoke and water will be discussed in the "A Vortex Particle Method for Smoke, Water, and Explosions."
(Courtesy of Andrew Selle, Stanford University; Nick Rasmussen, ILM; Ron Fedkiw, Stanford University and ILM)
How integral a role do the Papers play at SIGGRAPH and in the computer graphics industry in general?
MG: The SIGGRAPH Papers program has long been the finest international forum for disseminating groundbreaking, provocative, and important new work. The Papers program has served as a source of inspiration for generations of researchers, and it reaches out significantly into other programs. For instance, research results presented in the Papers program very often become integral parts of very successful courses in the SIGGRAPH Courses program. Also, some of the sketches of ongoing research eventually become complete Paper submissions, and some authors of systems-oriented work have presented prototypes of their innovations in the Emerging Technologies portion of SIGGRAPH.
The Papers program has also been a major source of knowledge and a driving force for the industry. Over the years, we have observed a continuous migration of the most recent research in image generation, modeling, and physics simulation into industrial, entertainment, and gaming applications. The games industry has become increasingly important for advanced visual and physics simulation.
The paper "Defocus Video Matting" discusses a technique for pulling mattes without a blue screen or special lighting. Instead, it compares video streams from differently-defocused cameras.
(Courtesy of Morgan McGuire et al., Brown University)
What themes or trends do you see regarding the Papers this year?
MG: There are a number of trends we can point to. The first one is that graphics researchers are doing more and more to bring reality into the computer. Complex lighting and shading models now become “data-driven”; that is, they are based on samples from the real world. This makes it possible to alter and simulate the appearance of human faces photorealistically. To this end, novel camera and acquisition devices have been created.
A second trend is the increasingly sophisticated simulation of physics. Various Papers deal with simulation of the complex interaction of media, such as liquids, smoke, or gas, and solid materials. Also, researchers have started to question the utility of triangle meshes for simulations and propose point-sampled models instead. Finally, various innovations make physics simulations interactive and real time on personal computers. This is very important for the development of more realistic games and will be supported by novel hardware architectures and processing units to be released by the industry very soon. In particular, novel chip generations and physics processing units will accelerate computations in computer games.
A third trend points toward advanced picture and video processing. We will see a variety of methods for panoramic stitching of videos, for making pictures three-dimensional, and for intelligent and user-friendly editing of video. Such methods might soon become tools people will utilize to edit their latest home videos.
The paper "Animating Gases with Hybrid Meshes" describes a technique for creating fluid simulations that can efficiently conform to arbitrary boundaries.
(Courtesy of Brian E. Feldman, James F. O'Brien, and Bryan M. Klinger of the University of California, Berkeley)
What selection criteria did the committee use when selecting Papers for acceptance?
MG: It is the tradition of the SIGGRAPH Papers program that technical excellence and innovation are the first and foremost selection criteria. We pick the submissions with the most exciting and most creative novel methods, the ones that stimulate future work by others, and the ones that constitute a tangible improvement over the state of the art. Equally important are technical soundness and a proof of concept. The authors have to provide a technically sound description of their innovation, and they have to clearly demonstrate — by comparison and analysis — that their methods work properly and are useful.
What challenges did you encounter when reviewing and choosing this year's Papers?
MG: The SIGGRAPH Papers program has a unique worldwide reputation as the premier forum for publishing outstanding technical contributions to computer graphics. The average quality of our submissions is therefore extremely high. It is generally very hard to select the subset of Papers for publications.
Over the past years, we have also faced a rapid growth in submission numbers. Retaining the quality of our well-established selection procedure has been a great challenge. We had to modify various aspects of the procedure to make it more efficient and flexible. Since the SIGGRAPH Proceedings are by far the most important publication for academic careers and promotions, we spent an enormous amount of effort to make sure the selection procedure was equally fair to everybody.
A new method for generating realistic 3D meshes of synthetic or real people from sparse marker data is presented in the paper "SCAPE: Shape Completion and Animation of People."
(Courtesy of Anguelov, Srinivasan, Koller, Thrun, and Rodgers of Stanford University; and James Davis of University of California at Santa Cruz)
Which Papers stand out in your mind as potential hits or must-attends for this year's show?
MG: That's difficult to say because in my opinion, all of them are outstanding contributions and advance the state of research in their respective field.
With that said, some of my recommendations include “Performance Relighting and Reflectance Transformation with Time-Multiplexed Illumination,” in which the authors present a novel system for facial appearance acquisition and modeling. This has the potential to become a very important tool for the movie industry.
I also recommend attending the session on fluid simulation, which will feature stunning animations. This session includes the Papers “Animating Gases with Hybrid Meshes,” “A Vortex Particle Method for Smoke, Water, and Explosions,” “Discontinuous Fluids,” and “Water Drops on Surfaces.”
An interesting approach to real-time deformations is presented in the Paper titled “Meshless Deformations Based on Shape Matchings.” In systems and hardware, promising novel graphics hardware architectures for ray tracing will be presented in the Paper titled “RPU: A Programmable Ray Processing Unit for Real-Time Ray Tracing,” and improved ray tracing algorithms will be introduced in the Paper, “Soft Shadow Volumes for Ray Tracing.”
I also recommend that attendees do not miss the session on video and image matting, which includes four outstanding Papers: “Defocus Video Matting,” Automatic Photo Pop-Up,” “Interactive Video Cutout,” and “Video Object Cut and Paste.” To learn the latest on human animation, I recommend the Paper, “SCAPE: Shape Completion and Animation of People.” For novel theoretical insights I recommend “A Frequency Analysis of Light Transport.”
Based on the work being presented this year, what predictions can you make about the kind of tools that will become commercially available in the near future?
MG: I expect that data-driven approaches, albeit technically complex and storage intensive, will eventually become commonplace in production environments for the visual simulation of complex surface phenomena. Also, I expect that the next generation of computer games will incorporate very complex, real-time physics simulations. Such simulations will be hardware-accelerated and will utilize methods and algorithms presented in the SIGGRAPH Papers Program.
Some innovations, such as raytracing hardware, might stimulate graphics hardware designers to rethink their traditional graphics pipelines. Finally, the video and image matting session points out the tools we will have at our disposal for advanced video editing on our PCs a few years down the road.
Any final thoughts?
MG: You can purchase the proceedings at SIGGRAPH 2005 and you can get access to individual Papers through the ACM digital library. Also, we encourage more of the nontechnical SIGGRAPH attendees to drop by and see some of the latest technical innovations. And finally, a fantastic and engaging video of the 2005 Papers will be shown in the SIGGRAPH Electronic Theater. Don't miss it!
Image courtesy of Andrew Selle, Stanford University; Nick Rasmussen, ILM; Ron Fedkiw, Stanford University and ILM)
SIGGRAPH 2005 Papers Advisory Board
Chair: Markus Gross
Swiss Federal Institute of Technology (ETH) in Zürich
Julie Dorsey
SIGGRAPH 2006 Papers Chair
Yale University (New Haven, CT)
Eugene Fiume
University of Toronto (Toronto, Canada)
John F. Hughes
Brown University (Providence, RI)
Session: "Skin & Faces"
8:30-10:15 a.m.
Session: "Hardware Rendering"
8:30-10:15 a.m.
Session: "Mesh Manipulation"
10:30 a.m.-12:15 p.m.
Session: "Illustration and Image-Based Modeling"
10:30 a.m.-12:15 p.m.
Session: "Meshes I"
3:45-5:30 p.m.
Session: "Video & Image Matting
3:45-5:30 p.m.
Tuesday, August 2
Session: "Meshes II"
8:30-10:15 a.m.
Session: "Perception"
8:30-10:15 a.m.
Session: "Motion Capture Data: Interaction and Selection"
10:30 a.m.-12:15 p.m.
Session: "Plants"
10:30 a.m.-12:15 p.m.
Session: "Capturing Reality I"
1:45-3:30 p.m.
Session: "Texture Synthesis"
1:45-3:30 p.m.
Session: "Capturing Reality II"
3:45-5:30 p.m.
Wednesday, August 3
Session: "Image Processing"
8:30-10:15 a.m.
Session: "Large Models & Large Displays"
8:30-10:15 a.m.
Session: "Fluid Simulation"
10:30 a.m.-12:15 p.m.
Session: "Reprise of UIST and I3D"
1:45-3:30 p.m.
Session: "Dynamics of Solids"
1:45-3:30 p.m.
Session: "Deformable Models"
3:45-5:30 p.m.
Thursday, August 4
Session: "Geometry on GPUs"
8:30-10:15 a.m.
Session: "Transperency & Translucency"
8:30-10:15 a.m.
Session: "Styles of Human Motion"
10:30 a.m.-12:15 p.m.
Session: "Appearance & Illumination"
10:30 a.m.-12:15 p.m.
Session: "Shape & Texture"
1:45-3:30 p.m.
Session: "Ray Tracing"
1:45-3:30 p.m.
Session: "Precomputed Light Transport"
3:45-5:30 p.m.
Continue the discussion on “Crosstalk” the Millimeter Forum.
