Final Project Proposal

I am proposing to use a Radiosity solution to the problem of view-independent global-illumination. This particular implementation will follow the notion of progressive refinement as proposed by Cohen, et al.

In terms of global illumination, radiosity is not new; literature regarding the subject is abundant. Regardless of this fact, it is still an exciting and active area of research, as most diffuse-diffuse light interations are solved using this method. However, regardless of its maturity, a complete radiosity solution still remains a daunting task.

I propose to recreate the classic cornell box using radiosity. The following comprise the key components:

· Scene generation – scene description is available from cornell

· Self adapting meshes

· Binary space partitioning and hierarchical bounding volumes for simple, fast VFC

· Progressive refinement approach for global illumination

· Hemicube approximation of form-factors

Again, radiosity is not novel, but its implementation has a nice intersection across the computer graphics domain.

The project deliverables will be:

· Win32 executable

· Full source code

o All code and tools will be original, no reuse of available resources, other than scene data