Experience using a dynamic adaptive hidden-surface removal algorithm for volume rendering within a MIMD architecture

Elapsed rendering times for all algorithm, designed to operate within a MIMD architecture, that accelerates the hidden-surface removal operation inside a volume visualization pipeline are presented. The algorithm is based on back-to-front and front-to-back hidden-surface removal/volume rendering algorithms that reduce image rendering time by adaptive rendering termination as soon as the image at a processor is informationally complete. Using these adaptively terminating algorithms, a technique for allowing each processor to dynamically and independently select the fastest volume rendering algorithm based on its local processing circumstances is formulated. Timing results obtained for the adaptively terminating back-to-front, adaptively terminating front-to-back, and dynamic hidden-surface removal/volume rendering algorithms for different volume sizes are presented. The results demonstrate that dynamic algorithm selection can significantly accelerate the volume rendering process