VI.3 - USE OF RESIDENCY MASKS AND OBJECT SPACE PARTITIONING TO ELIMINATE RAY-OBJECT INTERSECTION CALCULATIONS

Abstract

Publisher Summary Ray tracing has been a computationally intensive process. A large portion of the computation is used to compute the intersection of the sampling rays with the objects that make up the scene. This chapter explains the method of object-space partitioning in which the extent of the objects within the scene is divided into a number of smaller spaces or cells. Each cell is then assigned a list of objects that are contained within it. Partitioning of the object space can be done in a number of ways: uniform subdivision, octrees, and BSP trees. The sampling rays are then traversed from cell to cell until a satisfactory conclusion is reached in determining the visible surface for the ray. As each cell is processed, the ray is intersected with the list of objects that reside within the cell. That can result in unnecessary duplicated intersection calculations for objects that span multiple cells, for example—a long thin cylinder. The chapter further describes the Boolean properties of residency masks, which in combination with spatial partitioning can be used either to eliminate duplicated intersection calculations or as a selection mechanism for determining which objects in a scene must be examined. A residency mask is simply a bit vector in which each bit is assigned to a cell within the partitioned object space. A residency mask is assigned to each object defining the cells in which the given object resides.