I read Ch. 6 twice… Yes this part is abstruse.
1) Stocastic Relaxation Radiosity
Breadth first Monte Carlo iterative process. I love Jacob Iterative Method the most – by which, a image can be generated progressively. Isn’t it beautiful?
For curing image variance, Gathering for small patches, and Scattering for non-small patches.
2) Discrete Random Walk Radiosity
Comparing with #1, it is a depth first process, taking collision/survive/absortion test (Random Walk) as the stocastic estimation tool.
3) Photon Density Estimation Methods
It is a geometry free method, which I prefer the most. Unlike the above two those take mesh as the base, Photon-Density is a pixel based estimation method. But in essence I still consider it as a random walk process.
At last, several refinements as importance sampling is mentioned as one effective way for improving performance.
P.S. I would say this book is totally maths-based… good for researchers, but not so much for engineers..