Exploiting frame coherence with the temporal depth buffer in a distributed computing environment

We introduce the notion of a parallel temporal depth buffer (or t-buffer) to speed up the rendering process for ray traced animations. The t-buffer is a conceptually simple data structure which stores knowledge about coherent pixel values across the duration of an animation. Through the use of this information, we can potentially avoid recomputing a significant number of redundant pixel values and thus substantially reduce overall computing time. Although the t-buffer is general enough to be used with a variety of rendering methods, the algorithm employed in this work targets ray-traced animations and therefore readily lends itself to parallelization. A clustered (or distributed) computing environment is an efficient and effective candidate to handle the parallel computation due to its relatively low cost and the high computation to communication ratio inherent in the processing. The results from this method show substantial reductions in computation and overall rendering time with relatively low overhead.