A Distributed And Collaborative Graphics Rendering System

In this paper, we describe a distributed and collaborative graphics rendering system. This graphics rendering system integrates many hot techniques such as JavdWeb, VRML/graphics and distributed/parallel processing. Using this system, multi-users on the Internet can edit their 3D scenes in a collaborative manner. A distributed photo-realistic render is then used to generate the final image sequence. Summary JavdWeb technologies offer the basic infrastructure needed to integrate computers spread over the Internet into a seamless distributed computational resource [ 11. This Internet-based clustering has recently demonstrated its usefulness for several interesting and breaking-through applications. One of these applications is graphics rendering. The graphics rendering has been exploited to many areas ranging from scientific visualization to commercial entertainment. In this paper, we will present a distributed and collaborative graphics rendering system as shown in Figure 1. In this system, multiusers across the Internet can form a collaborative group and work together on editing three-dimensional scenes with the same interest. Our collaborative policies allow participants to dynamically register and leave a collaborative group. The VRML language is adopted in our implementation, since it is standard three-dimensional scene file format for Intemet applications [2]. Once the local VRML file is created, a user can either browse it locally or send it to the editor server for collaboration with other participants (see Figure 2). We propose mechanisms to enable synchronous and asynchronous edition of 3D scenes in a collaborative manner and to guarantee the correction of modification on shared scenes. Figure 3 (a) and (b) show a sequence of four-participant collaboration. To generate photo-realistic quality of images, we exploit a ray tracer. However, ray tracer is notorious for its high computational complexity [3]. To alleviate this problem, we design a distributed ray tracer exploiting sophisticated mechanisms to efficiently schedule aggregated computing power in the Intemet. For portability, we use Java /Web technologies to implement system GUI components. Figure 4 (a) and (b) show our GUI for the local editor and distributed ray tracer. We exploit PVM parallel programming technology [4] to develop our distributed ray tracer.