Evaluation of Subdivision Surfaces on Programmable Graphics Hardware

High-order smooth surface primitives, such as subdivision patches for example, are attractive for the modeling of free-form surfaces. In contrast to meshes they require only a few control points to specify large sections of a surface. Unfortunately, much of this bandwidth advantage is lost when such surfaces have to be tessellated on the CPU prior to transmission over the graphics bus and rendering on the graphics card. For surfaces built through linear combination of basis functions it is possible to precompute tessellations and use these to evaluate the surface at runtime in a simple computation performed entirely on a programmable graphics processor (GPU). The improved bandwidth requirements—only control points need transmission during animation, for example—coupled with the high performance of GPUs, allows us to achieve tessellation rates up to 24 million vertices per second on a 500 MHz GeForce FX. CR Categories: I.3.1 [Computer Graphics]: Hardware Architecture—Graphics processors; I.3.5 [Computer Graphics]: Computational Geometry and Object Modeling— Graphics processors;