Improved Indirect Photon Mapping with Weighted Importance Sampling

László Szirmay-Kalos, László Szécsi
Department of Control Engineering and Information Technology, Technical University of Budapest,
Budapest, Magyar Tudósok krt. 2, H-1117, HUNGARY


This paper offers a novel approach to the indirect photon mapping method. The placement of photons acting as virtual light sources is regarded as a cheap sampling scheme, allowing for the reuse of a complete shooting path at the cost a single shadow ray. In order to counter for its shortcomings, the variance reduction technique called weighted importance sampling is applied. This allows for the extension of the indirect photon mapping method for specular settings, because the weighting function can mimic surface BRDF characteristics disregarded by the virtual light source placement. On the other hand, weighted importance sampling also helps to eliminate the ``corner spikes'' caused by the fact that shooting cannot mimic the geometric factor of the connection rays. Advantages and problems are examined for several weighting schemes.