The presentation by Keven Villeneuve begins 34 minutes into the video. You can jump directly to that part of the video on YouTube.
Efficient Monte Carlo estimation of volumetric single scattering remains challenging due to various sources of variance, including transmittance, phase-function anisotropy, and geometric cosine foreshortening and squared-distance fall-off.
This presentation proposes several complementary techniques to importance sample each of these terms and their product. First, it introduces an extension to equiangular sampling to analytically account for the entire geometric term of point-normal emitters. It then includes transmittance and phase function via Taylor series expansion and/or warp composition. Scaling to complex mesh emitters is achieved through an adaptive tree splitting scheme, and multiple importance sampling combines the authors’ new samplers with existing approaches.
The techniques presented here consistently outperform state-of-the-art baselines on a diversity of settings.