Parallel depth merge: a paradigm for hidden surface removal

Visualization is a key aspect in geometric modeling. Since the processing required for realistic display is often considerable, the design of parallel visualization algorithms is trying task. In this paper a new parallel approach to Hidden Surface Removal, called Parallel Depth-Merge (PDM), is proposed. PDM, intrinsically parallel, couples any already known HSR object-space algorithm with an image-space hierarchical merger. Resulting data are returned in a format well suited both to be plotted on vector devices and to be used in shadow computations and shading algorithms. The proposed approach is designed for MIMD architectures, which use either shared or non shared memory, and does not require any special purpose hardware. The term "paradigm" is used to stress that the proposed approach is not based on the parallel redesigning of a known HSR algorithm. Conversely, it can be considered a general parallel solution to the problem, allowing different HSR object- space algorithms to be implemented in a uniform way on different MIMD architectures.