Raytracing is very computationally expensive and the classical solution is
to
employ a parallel architecture to break the task into n parts. This can mean
that:
the frames of an animated piece can be farmed out to and
that groups of scanlines can be rendered on different processors.
With lean overheads, the speed improvement can be close to n for n processors. The CAN raytrace system supports distributed network computation for multiple frame animation and also for single frames. The frame distribution is supported by a utility called "farmr" and the "-m" switch of the CAN raytracer enables intra-frame distribution. .sp 1 This document details how "farmr" and the "-m" switch are used to reduce the time taken for a raytraced image. It also gives examples of techniques that can be used to produce complex animation with minimal effort.
with images (6144286 bytes)
with no images (244044 bytes)