Connection Tracing for Photorealistic Rendering of Very Large and Complex Scenes

Research project Connection Tracing Eduard GRÖLLER 28.06.1999 In this project we propose to develop a new photorealistic rendering technique based on a new concept called `Connection Tracing`. The new method will make it possible to efficiently render very large scenes which contain a very large number of light sources and very difficult illumination settings. Such scenes cannot be efficiently handled by existing techniques. This is due to their methods, if any, for handling a large number of light sources and their problems of finding important paths of energy transport if they are close to singularities in path space. The basic features of Connection Tracing, that will successfully and efficiently solve the global illumination problem, even in very complex scenes are: - bidirectionality with eye and light paths - connection of eye and light paths and optimization by modification - incremental adaptive generation of eye and light paths instead of using delimited sequential passes - memory-efficient caching of eye and light paths - reuse of already generated eye and light paths - special adaptive determination of important light sources - no meshing We propose to develop new photorealistic rendering techniques based onconnection tracing that will provide: - high quality rendering - fast hardware based rendering - parallelization - stochastic full-spectral sampling of colours - support of multiple scattering participating media The new algorithm starts out by tracing a number of so-called light paths, paths of energy transport starting at light sources, and a number of so-called eye paths, paths of energy transport ending at the eye point of the observer in the scene. Initially these paths will not be connected. The new algorithm stores these paths based on the scattering points in the scene, and generates a number of paths connecting light sources with the eye point, by merging paths with scattering points that are very close to each other. This operation we call `connecting` paths. Due to highly specular reflection, the energy transported by connected paths can be significantly smaller as in the original paths that were not connected. Therefore the connected paths need to be iteratively modified to increase the amount of energy transported. Starting with this initial set of connected paths, the algorithm now incrementally adds new paths by spawning new paths at important scattering points. As new paths are generated their contribution to the overall energy transport is registered, and thus an image is generated.

Research in project P13600-N04 has resulted in several efficient methods for the simulation of global illumination, and for photorealistic and hardware accelerated rendering of complex scenes: - A photon map-based radiance estimation method that improves the quality of the global illumination solution in photon map global illumination simulation. - A particle map-based importance sampling technique which improves the performance of stochastic ray tracing- based rendering and global illumination simulation. - A hardware accelerated rendering method which allows to do interactive walkthroughs in globally illuminated glossy scenes. - An occlusion culling technique which allows to do interactive walkthroughs also in large scenes. Photon map global illumination simulation has proven to be a powerful method for ray tracing-based photorealistic rendering of globally illuminated scenes. Nevertheless, one of the weaknesses of this method has been that it uses a very coarse radiance estimation which may cause illumination artifacts. Our new photon map-based radiance estimation method avoids these illumination artifacts. This is done by taking the actual geometry of the illuminated surfaces into consideration. In stochastic ray tracing-based rendering and global illumination techniques a very large number of rays have to be shot into the scene to compute the global illumination solution and/or the final image. Our new importance sampling technique solves this problem by utilization of a particle map. The rays can therefore be precisely shot into directions with high contribution. Interactive walkthroughs in a globally illuminated static scene can be realized by doing the computationally expensive global illumination simulation in a preprocessing step. Our new method for interactive walkthroughs for soft glossy scenes solves this problem with directional light maps, which can be efficiently displayed with conventional graphics hardware. To achieve a real-time frame-rate for interactive walkthroughs in large scenes it is necessary to determine efficiently which objects are occluded, so that they can be culled. Our new conservative image-space occlusion culling method achieves this by utilization of a lazy occlusion grid that works efficiently with conventional graphics hardware.