Smart visible sets for networked virtual environments

Abstract

The real-time visualization of complex virtual environments across a network is a challenging problem in computer graphics. The use of pre-computed visibility associated to regions in space, such as in the potentially visible sets (PVS) approach, may reduce the amount of data sent across the network. However, PVS for a region may still be complex, and further partitions are necessary. In this paper we introduce the concept of a smart visible set (SVS), which corresponds to (1) a partition of PVS information into dynamic subsets that take into account client position, and (2) an ordering mechanism that enumerates these dynamic sets using a visual importance metric. Results comparing the SVS and PVS approach are presented.