Open Collections

Light-driven global illumination with a wavelet representation of light transport

University of British Columbia

This thesis considers the problem of global illumination: the modelling of light as it travels through a scene interacting with the objects contained within the scene. Starting with a description of the problem and a discussion of previous work, we explore a new approach called light-driven global illumination that offers several advantages over its predecessors: a lower asymptotic complexity, a wider range of representable surface interaction phenomena, and an absence of the need for "meshing" - object surface subdivision needed primarily to represent shadows. Light-driven global illumination is intermediate between local and global illumination. Representing light with wavelet basis functions, we are able to treat both the interaction between two surfaces and the interaction of a surface with a radiation field in a source-to-destination model that applies to whole surfaces, not just small elements. We have found this "wavelet radiative transfer" to be a valid way to generate and store complex global light field data as four-dimensional textures for incorporation in local illumination solutions. Wavelets can considerably reduce the otherwise substantial storage and reconstruction problems associated with doing this. We include several examples of this. We also discuss plausible illumination models, which are required to make light-driven global illumination work theoretically. Like wavelet radiative transfer, these models have application in other areas of rendering besides global illumination. Finally, we develop the theory behind light-driven global illumination and apply it successfully to some simple examples. While we find the algorithm to be quite slow compared to other well-known rendering algorithms, we analyze what is needed to make it competitive. In conclusion, we find that representing light with wavelets has a set of advantages that are independent of the comparative inefficiency of the light-driven algorithm.

Thesis/Dissertation

application/pdf

2009-06-22

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/9529

Computer Science

University of British Columbia

UBCV

DSpace