ABSTRACT
The representation of a 3D surface we have considered so far is limited to a polygon mesh. It has been widely used in real-time graphics, and as a matter of fact, it has been the only representation accepted by commodity graphics hardware. However, the polygon mesh is not a continuous accurate representation, but approximates a smooth surface. Such a nature reveals several problems. To list a few, its planar facets are not easy to hide in the rendered images, and we often need to control the levels of the representation detail [4]; if a high-resolution mesh is adopted to hide the polygonal nature, it may require large vertex/index buffers; editing or deforming a polygon mesh per se is a hard problem.
