Ray tracing with amalgams
Karl Kelvin Thompson · 1991
Traditional ray tracing databases that use bounding volumes attach only geometric properties (e.g. position, orientation, size) to each bounding volume. By also associating shading properties (e.g. color, translucency, reflectance) with each bounding volume, the database can become a multi-resolution database--a database that represents objects at different levels of detail. Specifically, the bounding region itself may be rendered to coarsely represent the objects inside it. With a minor extension to the traditional ray tracing algorithm, a ray tracing renderer can traverse such a database with fewer intersection tests per ray, improved antialiasing, and (arguably) better asymptotic time complexity. The enhanced database and algorithm are particularly suited to scenes containing a large number of primitives at a wide range of scales.