DISTRIBUTED AND COLLABORATIVE VISUALIZATION

Today's distributed systems can be harnessed to serve as cost-effective high-performance virtual machines for performing the large computations involved in volume rendering. The authors describe two algorithms for volume rendering by distributed ray-casting on a rectilinear mesh: One uses object space distribution, the other uses image space distribution. When implemented on a heterogeneous cluster of workstations on a network, these algorithms are able to produce translucent or shaded images of extremely large volume data sets. The algorithms and the facilities for collaborative visualization are implemented on top of the distribution and collaborative mechanisms of an environment called Shastra. Shastra tools are amenable to interoperation, and the Shastra environment provides facilities for media-rich interaction over a context of shared visualizations. The result is a synchronously conferenced collaborative visualization environment that lets users share large data sets, simultaneously view visualizations of data, and interact with multiple views while varying parameters.