Cornell University Program of Computer Graphics
Combining edges and points for interactive anti-aliased rendering.Kavita Bala, Bruce Walter, and Donald P. Greenberg.
Technical report PCG-02-3, Program of Computer Graphics, Cornell University, Jan 2002.
This paper presents a new rendering and display paradigm that uses both discontinuities (edges) and sparsely sampled shading (points) to interactively generate anti-aliased images of complex scenes. Geometric and shadow discontinuities in the image are found at interactive rates using a novel data structure, the Normal-Position Interval tree, and algorithms based on interval arithmetic. After projecting discontinuities on to the image plane, shading information is interpolated from nearby point samples called the edge-and-point image. An efficient interpolation algorithm uses this image to generate anti-aliased output images at interactive rates without using supersampling.
Our rendering technique is extensible, permitting the use of arbitrary shaders to collect radiance samples. Our software implementation supports interactive navigation and object manipulation in scenes that include complex lighting effects (such as global illumination) and geometrically complex objects. WE show that high-quality anti- aliased images of these scenes can be rendered at several frames per second on typical desktop machines.
This paper is available as a PDF file BWG02.pdf (519K).