Clustering-based hierarchical radiosity for dynamic environments

Abstract

This paper extends hierarchical radiosity and clustering techniques to dynamic environments, in which dynamic manipulations, such as repositioning an object and changing surface attributes, are repeatedly applied. The cluster techniques can be effective in reducing the computational complexity of both initial linking creation and link updating of dynamic manipulations. When the object is repositioned in a dynamic environment, the affected links, both energy and cluster links, can be identified rapidly by means of our proposed new strategy. As,the total net energy of the environment is decreased, we exploit the so-called history links to un-refine the affected patches. An effective progressive refinement strategy is also applied to further avoid the creation of unnecessary links.