Modeling, animating, and rendering complex scenes using volumetrictextures

摘要

Complex repetitive scenes containing forests, foliage, grass,nhair, or fur, are challenging for common modeling and rendering tools.nThe amount of data, the tediousness of modeling and animation tasks, andnthe cost of realistic rendering have caused such kind of scene to seenonly limited use even in high-end productions. The author describes hownthe use of volumetric textures is well suited to such scenes. Thesenprimitives can greatly simplify modeling and animation tasks. Morenimportantly, they can be very efficiently rendered using ray tracingnwith few aliasing artifacts. The main idea, initially introduced bynKajiya and Kay (1989), is to represent a pattern of 3D geometry in anreference volume, that is tiled over an underlying surface much like anregular 2D texture. In our contribution, the mapping is independent ofnthe mesh subdivision, the pattern can contain any kind of shape, and itnis prefiltered at different scales as for MIP-mapping. Although thenmodel encoding is volumetric, the rendering method differs greatly fromntraditional volume rendering. A volumetric texture only exists in thenneighborhood of a surface, and the repeated instances (called texels) ofnthe reference volume are spatially deformed. Furthermore, each voxel ofnthe reference volume contains a key feature which controls thenreflectance function that represents aggregate intravoxel geometry. Thisnallows for ray tracing of highly complex scenes with very few aliasingnartifacts, using a single ray per pixel (for the part of the scene usingnthe volumetric texture representation). The major technicalnconsiderations of our method lie in the ray-path determination and innthe specification of the reflectance function