A Zero-variance-based Sampling Scheme for Monte Carlo Subsurface Scattering

Jaroslav Křivánek
Charles University in Prague Eugene d'Eon
Weta Digital

We present a new variance reduction technique for unbiased Monte Carlo simulation of subsurface scattering—an important rendering problem for translucent materials like skin. The method is derived from the theory of zero-variance random walks, so far unexploited in computer graphics. Our method lowers variance per sample while increasing performance in Monte Carlo subsurface scattering simulation. The classical sampling results (left) use 100 samples/pixel. Our results use equal time with about 50% more samples/pixel (right).

Publication

Jaroslav Křivánek and Eugene d'Eon. A Zero-variance-based Sampling Scheme for Monte Carlo Subsurface Scattering. ACM SIGGRAPH 2014 Talks ... DOI | BibTeX

Links and Downloads

talk abstract	supplemental document	slide notes	slides

pdf (.3 MB)	pdf (2 MB)	pdf (4 MB)	pptx (3 MB)

Acknowledgments

Part of this work was done while the first author was with Weta Digital. This work was supported by the Czech Science Foundation (grant P202-13-26189S) and Charles University in Prague.