Context aware high-fidelity rendering over peer-to-peer systems

Abstract

High-fidelity rendering has witnessed widespread adoption in a number of disciplines and areas such as engineering, archaeology, and the entertainment industry with video games and special effects amongst others. Its pervasiveness is such that any improvements to the fundamental techniques employed will see the benefits propagate to these areas of application. This work focuses on the collaborative aspect of high-fidelity rendering where we demonstrate that peer-to-peer systems are a viable alternative to traditional client-server approaches. More specifically, issues of scalability that adversely affected previous work have been shown to be related to the rendering technique rather than the networking paradigm used for communication. Results show that in addressing the shortcomings of the irradiance cache, a marked improvement in system scalability and rendering performance has been achieved, close to two-fold in speed-up. Furthermore, the epidemiological nature of information propagation fails to prioritise content, precluding peers from receiving events in order of importance. To address this limitation and improve update propagation, a context-based dissemination approach was explored, wherein peers use meta-information to direct content updates. Results demonstrate that context-aware rendering does not incur any realisation penalties; a higher update frequency between peers sharing the same contexts suggests that the approach may be capable of providing more focused updates that possess greater relevance for the receiving peer.