Unbiased Photon Gathering for Light Transport Simulation

摘要

Photon mapping (PM) has been widely regarded as an efficient solutionrnfor light transport simulation, including challenging causticsrnpaths and many-bounce indirect lighting. The efficiency of PMrncomes from reusing traced photons. However, the handling of photonrngathering in existing PM algorithms is universally biased – thernexpected value of their results does not necessarily agree with therntrue solution of the rendering equation. We present a novel photonrngathering method to efficiently achieve unbiased rendering withrnphoton mapping. Instead of aggregating the gathered photons intornan estimated density as in classical photon mapping, we processrneach photon individually and connect the corresponding light subpathrnwith the eye sub-path that generates the gather point, creatingrnan unbiased path sample. The Monte Carlo estimate for such a pathrnsample is calculated by evaluating all relevant terms in a strict andrnunbiased way, leading to a self-contained unbiased sampling technique.rnWe further develop a set of multiple importance samplingrn(MIS) weights that allow our method to be optimally combined withrnbidirectional path tracing (BDPT), resulting in an unbiased renderingrnalgorithm that can efficiently handle a wide variety of light pathsrnand that compares favorably with previous algorithms. Experimentsrndemonstrate the efficacy and robustness of our method.