CAPSIM:Main
From Grand Forum
CAPSIM Mailing List
Project Name: From Capture to Simulation
Project Leader
Project Co-Leader
Researchers
- Eugene Fiume (Toronto)
- Wolfgang Heidrich (UBC)
- Paul Kry (McGill)
- Robert Bridson (UBC)
- Kyros Kutulakos (Toronto)
- Alla Sheffer (UBC)
Theme Distribution
Project Description
Over the past two decades, the computer graphics community has taken great strides in visually realistic, physics-based simulation of natural phenomena such as smoke, cloth, fluids (gases and liquids), etc. Some internationally recognized, key contributions have come from Canadian universities and industry. Unfortunately, while physical simulation can help us visualize phenomena we could not reproduce in real life (e.g. large explosions), the realistic simulation of many everyday phenomena is hitting a computational "brick wall". The main cause is the complexity, multitude, and interaction of the physical processes involved. Indeed, common events such as the lighting of a match, the burning of a sheet of paper or some other solid fuel, and the flow of water in a shallow, rocky stream are simply too complex to model computationally with today's methods. Similarly, simulations of human facial motions still suffer from unrealistic artifacts and fail to express the range of human emotions in a believable fashion. The key goal of this project is to tie together the capture (imaging) of natural phenomena with the physics-based simulation of such effects.
From the point of view of imaging, recent research has demonstrated that the visual realism of virtual environments can be improved significantly by capturing and analyzing data from the real world (e.g., photographs, 3D range scans, textures, and lighting models) (Ihrke, et al, 2008). Despite this progress, however, no attempts have been made to "import" these measurements into a simulation environment. Such a capability would enable interactive editing of previously-captured phenomena, would lead to creation of new simulation models, and would bring a new level of realism to existing systems.
This project will be the first systematic attempt world-wide to tackle such a tight integration of visual measurement and physical simulation. Research into new methods for capture, processing and mesh generation, and simulation will all be combined under a common project. A comprehensive treatment of this topic will be the subject of a long-term research program. Over the first several years the focus will be on individual phenomena that are of strong importance to both the game and animation industries. These include both volumetric effects such as fluids, fire, and smoke, as well as moving surfaces such as expressive faces or flowing garments.
Excellence of the Research
Development of Highly Qualified Personnel (HQP)
The project will support computer science graduate students, primarily from the University of British Columbia and the University of Toronto, which are home to Canada's top research groups in this area. In the past, many students working on precursors to this problem have gone to Canadian-based industry.
Networking and Partnerships
The team includes an outstanding group of Collaborating Researchers, such as Robert Bridson (UBC), Kiriakos Kutulakos (Toronto), and Alla Sheffer (UBC), who are expected to make major contributions to the project.
Knowledge and Technology Exchange and Exploitation
The project also has strong ties to Canadian industry, including companies such as Autodesk and 3D3 Solutions. Both Autodesk and 3D3 Solutions will both be active collaborators on the project, and will be able to make direct use of the methods developed in its context.
References to the Literature
I. Ihrke, K. Kutulakos, Hendrik P. A. Lensch, Marcus Magnor and Wolfgang Heidrich: State of the Art in Transparent and Specular Object Reconstruction, Eurographics STAR Report, 2008.
Publications
Presentations
|  |