Abstract
We present a practical method targeting dynamic shadow maps for many light sources in real-time rendering. We compute full-resolution shadow maps for a subset of lights, which we select with spatiotemporal reservoir resampling (ReSTIR). Our selection strategy automatically regenerates shadow maps for lights with the strongest contributions to pixels in the current camera view. The remaining lights are handled using imperfect shadow maps, which provide low-resolution shadow approximation. We significantly reduce the computation and storage compared to using all full-resolution shadow maps and substantially improve shadow quality compared to handling all lights with imperfect shadow maps.
Our new ReSTIR-sampled stochastic technique automatically selects and generates shadow maps for a few lights in each frame;
we handle other lights using imperfect shadow maps (ISMs). This allows more high-quality shadow-casting lights in raster engines without
resorting to low-quality ISMs. Notice that our method substantially improves the shadow quality both visually and numerically, as compared
to only ISMs and distance heuristics (where we pick closest lights to camera to generate shadow maps and other lights use a fixed visibility
value or ISMs). We also show the additional noise introduced by ReSTIR can be filtered by the denoiser [Schied et al. 2017]. For reference, we generate
shadow maps for all the lights and accumulate frames to remove the noise. This Emerald Square scene uses 35 omnidirectional lights (6
shadow maps for each) to approximate the emissive mesh lights.
Video
BibTeX
@article{Zhang2025,
author = {Song Zhang and Daqi Lin and Chris Wyman and Cem Yuksel},
title = {Many-Light Rendering Using ReSTIR-Sampled Shadow Maps},
journal = {Computer Graphics Forum (Proceedings of Eurographics 2025)},
year = {2025},
numpages = {12},
url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/cgf.70059},
doi = {10.1111/cgf.70059},
}