| Home > Publications > Reports > Informatics (CW) |
CW 267
Ph. Dutré, F. Suykens, Y.D. Willems
Optimized Monte Carlo path generation using genetic algorithms
Abstract
In this technical report we present a new method for optimizing the generation of paths in Monte Carlo global illumination rendering algorithms. Ray tracing, particle tracing, and bidirectional ray tracing all use random walks to estimate various fluxes in the scene. The probability density functions necessary to generate these random walks are optimized using a genetic algorithm, such that a significant reduction of the variance, and thus more reliable flux estimates, are obtained.
report.pdf / mailto: Ph. Dutre