Abstract: A method and apparatus for fast rendering of objects or scenes on a display controlled by a computer system. The scene to be rendered is stored as a series of triangle lists in an object data file. The triangle lists are presorted by the system into front- and back-facing triangles, as determined by scalar products of the triangle's normal vector with a vector from each respective triangle to the user's selected viewpoint. An array of pointers is generated for the front-facing triangles, and another array is generated for the back-facing triangles. The object or scene is rendered by first loading the attributes for the front-facing triangles to a frame buffer controller, then processing all of the front-facing triangles sequentially, without switching back and forth between front- and back-facing triangles. The attributes for the back-facing triangles are then loaded, and the back-facing triangles are then all processed sequentially.

Abstract: Various components of a color pixel stored in an accumulation buffer and respective components of a color pixel in a frame buffer are processed substantially simultaneously in partitioned multiplication operations and partitioned addition operations. An accumulation buffer weight is substantially multiplied by each component of the buffer pixel substantially simultaneously in a partitioned multiplication operation. The weighted buffer pixel is adjusted in some embodiments to effectively increase the range of accumulation buffer weights which can be effectively processed by the processor. For example, the weighted buffer pixel is doubled to effectively extend the effective range of the accumulation buffer weight from approximately -0.5 to 0.5 to at least the range of approximately 0.0 to approximately 1.0.