Abstract
The authors investigate four different concepts which are candidates to resolve the hidden surface removal problem without the use of priorities. All four are based on determining a linear depth order consistent with the partial order of primitives in a span. A span is a small subimage of the image being generated. The first approach examines parallel sorting of all the depth values of all primitives in a span. The second approach, for each pixel in a span, builds a list of all the primitives in that span, totally ordered by depth. These lists represent the partial order of the primitives, and are later enumerated to give a total order. The third approach performs up to k pairwise comparisons in parallel between primitives to determine if a depth relationship exists. If one exists, the pair is stored in a partial order graph which is later enumerated to create a total order. The fourth approach involves the use of a pipeline of length k, which, at any given time, contains up to k primitives which are closest. If there are more than k primitives in a span, more than one iteration is needed. The fourth approach appears to be the most promising, even though there are situations in which it generates an incorrect ordering. >
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have