Home | english  | Impressum | Datenschutz | Sitemap | KIT

Selective guided sampling with complete light transport paths

Florian Reibold, Johannes Hanika, Alisa Jung, and Carsten Dachsbacher

To appear in ACM Transactions on Graphics 37(6) (Proceedings of SIGGRAPH Asia 2018)

Our (filtered) Insets Unguided contribution
Guided contribution
A dining room lit from the outside by an area light and by a spot light on the table. The insets show that our (unidirectional) guided sampling is able to resolve the caustic much better than path tracing (PT) in the same time. Guided sampling learns the difficult parts of the light transport in 3 minutes (guided contribution, bottom right image) and rendered for 27 minutes. Our approach is selective, i.e. the guided sampler automatically focuses only on the parts of the illumination which are poorly sampled by the unguided estimator. The images on the right show the unguided and guided component of our method.

Abstract

Finding good global importance sampling strategies for Monte Carlo light transport is challenging. While estimators using local methods (such as BSDF sampling or next event estimation) often work well in the majority of a scene, small regions in path space can be sampled insufficiently (e.g. a reflected caustic). We propose a novel data-driven guided sampling method which selectively adapts to such problematic regions and complements the unguided estimator. It is based on complete transport paths, i.e. is able to resolve the correlation due to BSDFs and free flight distances in participating media. It is conceptuall simple and places anisotropic truncated Gaussian distributions around guide paths to reconstruct a continuous probability density function (guided PDF). Guide paths are iteratively sampled from the guided as well as the unguided PDF and only recorded if they cause high variance in the current estimator. While plain Monte Carlo samples paths independently and Markov chain-based methods perturb a single current sample, we determine the reconstruction kernels by a set of neighbouring paths. This enables local exploration of the integrand without detailed balance constraints or the need for analytic derivatives. We show that our method can decompose the path space into a region that is well sampled by the unguided estimator and one that is handled by the new guided sampler. In realistic scenarios, we show 4x speedups over the unguided sampler.

Downloads

to appear