Tracking data structures coherency in animated ray tracing for real-time 3D-rendering

Ray tracing is a well known method for photorealistic image synthesis, volume visualization and rendering. Over the last decade the method is being adopted throughout the research community around the world. With the advent of the high speed processing units, the method has been emerging from offline rendering towards real time rendering. The success behind ray tracing algorithms lies in the use of acceleration data structures and modern processing power of CPUs and GPUs. kd-tree is one of the most widely used data structures based on surface area heuristics (SAH). The major bottleneck in kd-tree construction is the time consumed to find optimum split locations. In this paper, we propose a prediction algorithm for animated ray tracing based on Kalman filtering. The algorithm successfully predicts the split locations for the next consecutive frame in the animation sequence. Thus, giving good initial starting points for one dimensional search algorithms to find optimum split locations - in our case parabolic interpolation combined with golden section search. With our technique implemented, we have reduced the ldquorunning kd-tree constructionrdquo time by between 78% and 87% for dynamic scenes with 16.8 K and 252 K polygons respectively.