On comparing ray shooting algorithms

In this paper we discuss a methodology for comparing various ray shooting algorithms through a set of experiments performed on a set of scenes. We develop a computational model for ray shooting algorithms, which allows us to map any particular ray shooting algorithm to the computational model. Further, we develop a performance model for ray shooting algorithms, which establishes the correspondence between the computational model and the running time of the ray shooting algorithm for a sequence of ray shooting queries. Based on these computational and performance models, we propose a set of parameters describing the use of a ray shooting algorithm in applications. These parameters allows us to make a fair comparison of various ray shooting algorithms for the same set of input data, i.e., the same scene and the same sequence of ray shooting queries, but virtually independently of hardware and implementation issues. Under certain conditions, the proposed comparison methodology enables cross-comparison of published research work without reimplementation of other ray shooting algorithms. r 2003 Elsevier Ltd. All rights reserved.