The mapping of images onto surfaces may substantially increase the realism and information content of computer-generated imagery. The projection of a flat source image onto a curved surface may involve sampling difficulties, however, which are compounded as the view of the surface changes. As the projected scale of the surface increases, interpolation between the original samples of the source image is necessary; as the scale is reduced, approximation of multiple samples in the source is required. Thus a constantly changing sampling window of view-dependent shape must traverse the source image. To reduce the computation implied by these requirements, a set of prefiltered source images may be created. This approach can be applied to particular advantage in animation, where a large number of frames using the same source image must be generated. This paper advances a "pyramidal parametric " prefiltering and sampling geometry which minimizes aliasing effects and assures continuity within and between target images. Although the mapping of texture onto surfaces is an excellent example of the process and provided the original motivation for its development, pyramidal parametric data structures admit of wider application. The aliasing of not only surface texture, but also highlights and even the surface representations themselves, may be minimized by pyramidal parametric means.

Traditional graphics hardware architectures implement what we call the push architecture for texture mapping. Local memory is dedicated to the accelerator for fast local retrieval of texture during rasterization, and the application is responsible for managing this memory. The push architecture has a bandwidth advantage, but disadvantages of limited texture capacity, escalation of accelerator memory requirements (and therefore cost), and poor memory utilization. The push architecture also requires the programmer to solve the binpacking problem of managing accelerator memory each frame. More recently graphics hardware on PC-class machines has moved to an implementation of what we call the pull architecture. Texture is stored in system memory and downloaded by the accelerator as needed. The pull architecture has advantages of texture capacity, stems the escalation of accelerator memory requirements, and has good memory utilization. It also frees the programmer from accelerator texture memory management. However, the pull architecture suffers escalating requirements for bandwidth from main memory to the accelerator. In this paper we propose multi-level texture caching to provide the accelerator with the bandwidth advantages of the push architecture combined with the capacity advantages of the pull architecture. We have studied the feasibility of 2-level caching and found the following: (1) significant re-use of texture between frames; (2) L2 caching requires significantly less memory than the push architecture; (3) L2 caching requires significantly less bandwidth from host memory than the pull architecture; (4) L2 caching enables implementation of smaller L1 caches that would otherwise bandwidth-limit accelerators on the workloads in this paper. Results suggest that an L2 ...

This dissertation explores the use of an analytic visibility algorithm for the high-speed generation of anti-aliased images with textures. Wben visibility is known analytically before any sampling takes place, low-pass filtering for anti-aliasing can be done to arbitrary accuracy at a cost proportionate to the output resolution. Furthermore, since the filtering and sampling processes can be expressed as a set of integrations over the image plane, the filtering process can be decomposed into a set of sums of integrations over each visible surface in the image plane, allowing the rendering of each visible surface to be done in parallel using an image buffer to accumulate the results. Thus, analytic visibility can serve as the basis for high-speed, high-quality image synthesis. In this dissertation, algorithms for computing analytic visibility and for producing filtered renderings from the resulting visible surfaces are presented. In order to provide real-time performance, these algorithms have been