Affiliation:
1. The University of North Carolina at Chapel Hill
Abstract
We demonstrate an efficient method of rendering alias-free synthetic images using precomputed convolution integrals. The method is based on the observation that a
visible polygon fragment's contribution
to an
image is
solely a function of its position and shape, and that within a reasonable level of accuracy, a limited number of shapes represent the majority of cases encountered in
images commonly rendered
.The convolution integral is precomputed for all pixels affected by the polygon fragment and is stored in a table. Completely visible fragments which are either triangular or trapezoidal produce two indices into the table. Most other fragments which are represented as differences of simple fragments. The remaining cases are characterized by a bit-mask for which each bit has a corresponding set of look up tables.The basic technique has been applied to several fundamentally different rendering algorithms. In addition, we illustrate a version of the newly introduced nonuniform sampling technique implemented in the same program, but with different table values.
Publisher
Association for Computing Machinery (ACM)
Subject
Computer Graphics and Computer-Aided Design,General Computer Science
Cited by
17 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献
1. Forward rasterization;ACM Transactions on Graphics;2006-04
2. Quadrature prefiltering for high quality antialiasing;ACM Transactions on Graphics;1996-10
3. Tracing rays with the area sampling machine;The Visual Computer;1995-09
4. An adaptive supersampling method;Lecture Notes in Computer Science;1995
5. Volume Synthesis Principles;Computer Graphics: Systems and Applications;1993