Many applications of computer graphics require accurate recreations of the appearance of the physical world. For example, computer-generated images used for product design need to be as visually close as possible to the real-world products. Applications in entertainment industry such as movie production and video games increasingly demand photorealism of computer-generated images. Computer simulation of how light interacts with given virtual shapes and materials, or light transport simulation, is a natural approach to achieve such photorealism.
In this talk, I will describe a new light transport simulation framework that significantly expands a range of optical configurations that we can simulate. The framework builds on a novel density estimation method, called progressive density estimation, which addresses fundamental limitations of existing density estimation methods. Using progressive density estimation, we can simulate many optical configurations that are simply impractical to handle with any other existing algorithm. In particular, this framework provides the first and currently the only way to efficiently simulate complex lighting fixtures from the filament/LED-level.
