Just implemented Jensen and Christensen(1998)'s approach of computing multiple scattering of participating media in PBRT.
This is the sponza scene with homogeneous media when only single scattering is accounted. Because there is not much direct lighting, the media is hardly to see.
This is the same scene as above but when both single
and multiple scattering are accounted. Now the presence of media is obvious, thanks to the light multiple scattering within the media. This is actually a kind of global illumination for volume effect.
Update: Fixed some bugs and tested on a scene with volume caustics.
Single Scattering
With multiple scattering, volume caustics appeared. Small steps needed to get volume caustics correct.
