Efficiency-aware multiple importance sampling for bidirectional rendering algorithms

SIGGRAPH 2022 (journal)

Two scenes that are challenging to render efficiently without user guidance: The caustics and strong indirect illumination in the Fish scene (left) resolve much faster with a bidirectional method such as VCM. But the simpler Target Practice scene (right) renders 2x slower with VCM than with forward path tracing. Our method renders both scenes efficiently by automatically setting the number of light subpaths to trace, the number of bidirectional connections to make, and in which pixels to perform photon density estimation (the 'merge mask' specifies a probability for performing a photon lookup in a pixel).

Abstract

Multiple importance sampling (MIS) is an indispensable tool in light-transport simulation. It enables robust Monte Carlo integration by combining samples from several techniques. However, it is well understood that such a combination is not always more efficient than using a single sampling technique. Thus a major criticism of complex combined estimators, such as bidirectional path tracing, is that they can be significantly less efficient on common scenes than simpler algorithms like forward path tracing. We propose a general method to improve MIS efficiency: By cheaply estimating the efficiencies of various technique and sample-count combinations, we can pick the best one. The key ingredient is a numerically robust and efficient scheme that uses the samples of one MIS combination to compute the efficiency of multiple other combinations. For example, we can run forward path tracing and use its samples to decide which subset of VCM to enable, and at what sampling rates. The sample count for each technique can be controlled per-pixel or globally. Applied to VCM, our approach enables robust rendering of complex scenes with caustics, without compromising efficiency on simpler scenes.

Downloads and links

Media

Fast-forward video

Talk video

BibTeX reference

@article{Grittmann:2022:EfficiencyAwareMIS,
  author = {Pascal Grittmann and \"Omercan Yazici and Iliyan Georgiev and Philipp Slusallek},
  title = {Efficiency-aware multiple importance sampling for bidirectional rendering algorithms},
  journal = {ACM Transactions on Graphics (Proceedings of SIGGRAPH)},
  year = {2022},
  volume = {41},
  number = {4},
  doi = {10.1145/3528223.3530126}
}