| 摘要 |
Both geometry and detail of an object's surface are simultaneously analyzed so as to generate a minimal set of data necessary to display the object with minimal visible errors at maximum drawing speed. All knowledge about the surface from high-level definitions (such as triangles, non-uniform rational B-splines (NURBS), textures, etc.) reduced into an internal form (a surface map) is (i) optimized according to visibility rules and (ii) output as a minimized set of machine code (triangles with position, color and normals) to a processor, e.g., a specific graphic processor. Shapes and groups (areas of common lighting and material type) representing an object are first identified. Data in each group is then re-sampled as a vector-valued field in a parametric U and V grid space. The surface map is then optimized by removing data points in both the U and V directions based upon a visibility function that determines which surface data most substantially contributes to the perceived distinctiveness of the representation. After removing unnecessary data points, the points are reconnected to form polygons, which are then split into a set of triangles used to display the surface at high speed with minimal visible errors over a variety of scale factors.
|
