Using KD-trees to guide bounding volume hierarchies for ray tracing

Ray tracing is a simple yet powerful technique for investigating complex environments and a standard method of accelerating the ray casting process has involved constructing bounding volume hierarchies which have required either the use of precalculated costs or creating them as part of the model construction. We propose a method using the kd-tree which will guide the construction of a bounding volume hierarchy as well as providing a method for its traversal by individual rays. By taking advantage of the tree structure's recursive nature we show how to construct an algorithm with very good performance characteristics which in typical real life cases is much more computationally efficient than previous methods. Results are presented using some standard ray tracing benchmarks.