Real-time All-frequency Global Illumination with Radiance Caching

Youxin Xing, Gaole Pan, Xiang Chen, Ji Wu, Lu Wang, Beibei Wang
CVMJ 2023, to appear

GI
Results of ZeroDay using our method. The comparison between our final result and the reference is shown on the left. The result of the convolution of BRDFs and lighting caching is shown on the right (top). The ray tracing result is shown on the right (bottom).

Abstract

Global Illumination (GI) plays a crucial role in rendering realistic results for the virtual exhibition, e.g., the virtual car exhibition. These scenarios usually include all-frequency bidirectional reflectance distribution functions (BRDFs), although the geometry and the light configuration might be static. Rendering all-frequency BRDFs in real-time is still challenging due to the complex light transport. Existing approaches, including precomputed radiance transfer, light probes, or the most recent path tracing-based approaches (ReSTIR PT), can not satisfy both quality and performance requirements at the same time. In this paper, we propose a practical hybrid global illumination approach, combining ray tracing and cached GI by caching the incoming radiance with wavelets. Our approach can produce close results to offline renderers at the cost of only about 17 ms at runtime and is robust over all-frequency BRDFs. Our approach is designed for applications with static lighting and geometries, like the virtual exhibition.

Downloads

PDF format Video

BibTex

@article{Xing:2023:RealtimeGI,
  author = {Youxin Xing and Gaole Pan and Xiang Chen and Ji Wu and Lu Wang and Beibei Wang},
  title = {Real-time All-frequency Global Illumination with Radiance Caching},
  journal ={Computational Visual Media},
  year = {2023},
}

Back to Homepage