Plane Sampling for Light Paths from the Environment Map

Abstract

We present a method to start light paths on an emitting environment map in the context of Monte Carlo global illumination with image-based lighting or physical sky models. With this technique, we can efficiently render unbiased caustics from these kinds of lights, for example using bi-directional path tracing. Additionally, it is now possible to use algorithms like multiple importance sampling, photon mapping, and instant radiosity correctly with environment maps.