Time‐critical rendering algorithm with incorporation of LoD, visibility culling and object impostor

Given enough CPU time, present graphics technology can render near‐photorealistic images. However, for real‐time graphics applications such as virtual reality systems, developers must make explicit programming decisions, trading off rendering quality for interactive update rates. In this papar we present a new algorithm for rendering complex 3D models at near‐interactive rates that can be used in virtual environments composed of static or dynamic scenes. The algorithm integretes the techniques of level of details (LoD), visibility computation and object impostor. The method is more suitable for very dynamic scenes with high depth complexity. We introduce a new criterion to identify the occluder and the occludee: the object that can be replaced by its LoD model and the one that can be replaced by its impostor. The efficiency of our algorithm is then illustrated by experimental results.