Realtime Ray Tracing and its use for Interactive Global Illumination
Ingo Wald, Timothy J. Purcell (#), Joerg Schmittler, Carsten Benthin and Philipp Slusallek
( Computer Graphics Group, Saarland University; '#'= Computer Graphics Group, Stanford University)
Eurographics State of the Art Reports, 2003
Research on realtime ray tracing has recently made tremendous advances. Algorithmic improvements together with optimized software implementations already allow for interactive frame rates even on a single desktop PC. Furthermore, recent research has demonstrated several options for realizing realtime ray tracing on different hardware platforms, e.g. via streaming computation on modern graphics processors (GPUs) or via the use of dedicated ray tracing chips. Together, these developments indicate that realtime ray tracing might indeed become a reality and widely available in the near future.
As most of todays global illumination algorithms heavily rely on ray tracing, this availability of fast ray tracing technology creates the potential to finally compute even global illumination – the physically correct simulation of light transport – at interactive rates.
In this STAR, we will first cover the different research activities for realizing realtime ray tracing on different hardware architectures – ranging from shared memory systems, over PC clusters, programmable GPUs, to custom ray tracing hardware. Based on this overview, we discuss some of the advanced issues, such as support for dynamic scenes and designs for a suitable ray tracing API. The third part of this STAR then builds on top of these techniques by presenting algorithms for interactive global illumination in complex and dynamic scenes that may contain large numbers of light sources. We believe that the improved quality and the increased realism that global illumination adds to interactive environments makes it a potential “killer application” for future 3D graphics.