Gradient-domain Path Tracing

Marco Manzi1 Markus Kettunen2 Jaakko Lehtinen3 Fredo Durand4 Matthias Zwicker1
1University of Bern 2Aalto University 3Aalto University and Nvidia 4Massachusetts Institute of Technology

In ACM Transactions on Graphics (Proceedings of SIGGRAPH), 34(4), August 2015

Teaser
Comparing gradient-domain path tracing (G-PT, L1 reconstruction) to path tracing at equal rendering time (2 hours). In this time, G-PT draws about 2,000 samples per pixel and the path tracer about 5,000. G-PT consistently outperforms path tracing, with the rare exception of some highly specular objects. Our frequency analysis explains why G-PT outperforms conventional path tracing.

Abstract

We introduce gradient-domain rendering for Monte Carlo image synthesis. While previous gradient-domain Metropolis Light Transport sought to distribute more samples in areas of high gradients, we show, in contrast, that estimating image gradients is also possible using standard (non-Metropolis) Monte Carlo algorithms, and furthermore, that even without changing the sample distribution, this often leads to significant error reduction. This broadens the applicability of gradient rendering considerably. To gain insight into the conditions under which gradient-domain sampling is beneficial, we present a frequency analysis that compares Monte Carlo sampling of gradients followed by Poisson reconstruction to traditional Monte Carlo sampling. Finally, we describe Gradient-Domain Path Tracing (G-PT), a relatively simple modification of the standard path tracing algorithm that can yield far superior results.

Additional Information