Acceleration techniques for progressive refinement radiosity

The progressive refinement radiosity method provides a means for computing the global illumination of an environment in time proportional to the size of the environment. Additionally, the method supplies a sequence of solutions that monotonically converge to a high-quality result, allowing it to be used for interactive applications that deliver "near real-time" updates as the computation proceeds. This paper presents two methods for accelerating the radiosity method. First, the traditional hemi-cube algorithm is modified to accelerate the expensive form-factor calculation. Second, the radiosity method is parallelized across a coarse-grain network, and the efficiency of the parallel system is discussed.