Abstract
Ray tracing is a simple, yet powerful, image synthesis technique that can produce realistic images. Unfortunately, ray tracing creates several artifacts because of the mathematical properties of a ray. A new technique called adaptive cone tracing, a combination of cone tracing and adaptive supersampling, is introduced to enhance the realism of ray traced images. Spatial aliasing is reduced by tracing a cone surrounding the viewpoint and pixel. When the boundary of an object intersects this cone, the cone is recursively subdivided to determine the illumination of the pixel to any level of accuracy. Adaptive cone tracing will always detect a projection of an object onto a pixel, so that small or thin objects will not be missed. It will also detect when there is no aliasing from the boundaries of objects. Finally, adaptive cone tracing can simulate the effects of jittered sampling near the boundaries of objects. Spatial aliasing of the edges of shadows from point light sources is also reduced by tracing additional shadow rays, not eye rays. Further, additional shadow rays only have to test the objects causing the shadow. Finally, adaptive cone tracing can simulate the effects of jittered sampling around the edges of shadows. Distributed (or area) light sources can be modeled by forming a cone towards the light (a line, rectangle, sphere or arbitrary polygon) and subdividing in the same manner as above. The resulting penumbrae can be calculated to any level of accuracy and non-penumbral regions only require one shadow cone test. Further, additional shadow sub-cones only test the objects causing the shadow. Finally, adaptive cone tracing can also simulate the effects of jittered sampling in the penumbral regions. Aliasing around the edges of reflections is reduced by detecting the boundaries of reflections and tracing additional reflected rays, not eye rays. Reflections can also be made fuzzy by de-focusing the reflected cone. Objects can have sharp or fuzzy reflections and the amount of fuzziness is maintained as a surface property. In both cases, adaptive cone tracing can simulate the effects of jittered sampling near the boundaries of reflections.
Genetti, Jon Dudley (1993). Image synthesis with adaptive cone tracing. Texas A&M University. Texas A&M University. Libraries. Available electronically from
https : / /hdl .handle .net /1969 .1 /DISSERTATIONS -1518984.