Collaboration-based global illumination for IlluminaPRT

Abstract

In the field of computer graphics, especially high-fidelity rendering, shared-memory parallelism has garnered a lot of interest due to the generally embarrassingly parallel nature of rendering computations. Distributed systems have also been instrumental in advancing the quality, complexity and fidelity of image synthesis, mostly through the use of the master-worker paradigm for a single client. Notwithstanding, collaborative rendering, where multiple independent clients (observers) are visualising and possibly interacting on a single scene and share the results of common computations they might have carried out, has yet eluded mainstream research. In this final year project, we investigate collaborative high-fidelity rendering between multiple clients; specifically, we attempt to demonstrate that collaborative rendering may provide benefits in terms of improved image quality and reduced computation times.