Final radiosity gather step using a Monte Carlo technique with optimal importance sampling

In this paper, an improved Monte Carlo based technique is presented to perform a per-pixel final gather step, also referred to as a local pass in radiosity computations. Using importance sampling based on the results from the finite-element radiosity computation, we achieve a very high image quality at a much lower cost compared to other common techniques. The theory in this paper examplifies how finite element and Monte Carlo techniques for rendering can be combined elegantly.