| 摘要 |
In rendering an image including a set of pixels (41) and representing a 3D scene including surface elements, each pixel (41ij) is constructed from light contributions corresponding to rays coupling the pixel and the surface elements in function of their light interaction features. Memory elements (430ij, 431ij...) are respectively associated with the pixels, storing data on ray directions (V0...V7). The rays are ordered in each memory element for one or several surface elements, by an ordering method applied to the rays in function of the ray directions in the memory element. The ordered rays are available to computation processing circuits in processing the pixels. This makes possible a significant reduction of cache misses in parallel computing. |
