Variance Based Go with the Winner Strategy in Path Tracing

Szirmay-Kalos, L., Antal, Gy., Sbert, M.

This paper proposes a new random walk strategy that minimizes the
variance of the estimate using statistical estimations of local
and global features of the scene. Based on the local and global
properties, the algorithm decides at each point whether a
Russian-roulette like random termination is worth performing, or
on the contrary, we should split the path into several child
paths. In this sense the algorithm is similar to the
go-with-the-winner strategy invented in the general Monte-Carlo
context. However, instead of establishing thresholds to make
decisions, we compute the number of child paths on a continuous
level and show that Russian roulette can be interpreted as a kind
of splitting using fractional number of children. The new method is built into
a path tracing algorithm. Comparing it with the classical path tracing approach
we concluded that the new method reduced the variance significantly.