Rendering post production applications in distributed environments

The paper considers the problem of rendering post production applications in a distributed environment. The aim is to limit the concessions that conventional, in-house methods force graphics operators to make to the quality of the end products to get them on time and within budget. A reference architecture for distributed rendering is proposed that takes advantage of the parallel properties of the application to reduce the job turnaround time. Performance issues that stimulate the industrial sector to use the environment are discussed through a simplified evaluation framework. The authors also examine how to serve multiple customers in a metacomputing environment employing limited resources. User strategies for submitting jobs over a priority-based service system that minimize the associated rendering cost are defined. Furthermore, a pricing scheme is implemented that succeeds in driving the user actions to an operating point in which certain provider objectives are met and, at the same time, discourages the users from utilizing a congested system, in which meeting of deadlines cannot be assured. Results and directions are not only restricted to the post production area, but apply over a wide range of remote operations.