A general solution requires integration of the global illumination method within a more general rendering system , such as Performer – CAVELib

This article presents a solution for the rendering of immersive VR using real-time global illumination. One of the important reasons for using VR in an application is that participants should respond realistically to virtual objects and events—for example, in applications concerned with training or rehabilitation. Therefore, investigating the factors that might be critical in producing such realistic responses is an important area of scientific investigation, with implications for the engineering of successful VR applications. An obvious factor to consider is the realism of the illumination. While a static globally illuminated scene is well within the grasp of current methods, it places limits on the range of environments and tasks that can be effectively represented. When scenes are rendered with global illumination, they not only look more realistic but also critically have dynamic shadows and reflections of objects, most importantly of the virtual body of the participant interacting in the VR. In other words, as participants move through the VR they would see real-time correlations of their activities through not only changing object reflections but also dynamic changes of the shadows and reflections of their body. This can have an anchoring effect that profoundly situates the participant within the virtual environment. The problems faced in rendering with real-time global illumination, however are twofold: First, the computational complexity of such a real-time rendering system while achieving an acceptable frame rate is daunting. Second, we are faced with a choice of how this is to be achieved—one option is to construct an ad hoc system with support for related tasks such as tracking, display management, and synchronization. Alternatively, a general solution requires integration of the global illumination method within a more general rendering system, such as Performer–CAVELib (an API for the Cave Automatic Virtual Environment), DIVE (Distributed Interactive Virtual Environment), or XVR (Extreme VR), which can again be a complex task. In this article, we present the Virtual Light Field (VLF) paradigm as a solution to this problem. We discuss its potential and advantages in such an application. Finally, we present details of our integration of the VLF rendering method within XVR to provide a practical real-time global illumination solution.