Jaroslav Křivánek

Survey of Markov Chain Monte Carlo Methods
in Light Transport Simulation

Martin Šik Charles University, Prague Render Legion, a.s.

Jaroslav Křivánek Charles University, Prague Render Legion, a.s.

Abstract

Two decades have passed since the introduction of Markov chain Monte Carlo (MCMC) into light transport simulation by Veach and Guibas, and numerous follow-up works have been published since then. However, up until now no survey has attempted to cover the majority of these methods. The aim of this paper is therefore to offer a first comprehensive survey of MCMC algorithms for light transport simulation. The methods presented in this paper are categorized by their objectives and properties, while we point out their strengths and weaknesses. We discuss how the methods handle the main issues of MCMC and how they could be combined or improved in the near future. To make the paper suitable for readers unacquainted with MCMC methods, we include an introduction to general MCMC and its demonstration on a simple example.

Reference

Martin Šik and Jaroslav Křivánek. Survey of Markov Chain Monte Carlo Methods in Light Transport Simulation. IEEE Transactions on Visualization and Computer Graphics, 2018
DOI | BibTeX

Links and Downloads

paper fulltext	supplemental document
pdf (17 MB)	zip (0.2 MB)

Acknowledgments

The authors thank W. Jakob for his renderer Mitsuba, which was used to generate most of the images. We also thank M. Pharr, J. Hanika, A. S. Kaplanyan, T.-M. Li, B. Segovia, H. Otsu, T. Hachisuka, W. Jakob, and J. Lehtinen for permission to use their examples/scenes/images/source code. This work was supported by the grant SVV–2017–260452 and by the Czech Science Foundation grant 16–18964S.