arXiv · 2611.12103
Real-Time Volumetric Light Transport Inference without a Path
Abstract
We present a real-time method for reconstructing fully integrated pixel radiance in rendered volumetric media under direct and ambient environmental illumination, using compact, low-cost scene features and statistics. Unlike Monte Carlo denoising pipelines for volumetric rendering, which depend on computationally expensive path-traced radiance estimates, our approach directly infers the final light-transport solution from inexpensive primary-ray and environment sample evaluations. We show that these features form a near-sufficient representation of the scene, enabling accurate reconstruction at substantially lower computational cost. Using an attention-augmented image-to-image network, we demonstrate that complete pixel radiance can be inferred directly from these auxiliary features, without requiring the original radiance signal. Operating in image space also provides a more stable computational cost as the camera zooms into or out of the volume, with substantially less dependence on volume density than approaches based on explicit multiple-scattering simulation. We demonstrate real-time performance and high reconstruction accuracy in challenging volumetric cloud scenarios.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Ludwic Leonard, Ruediger Westermann. 2026-11-09. Real-Time Volumetric Light Transport Inference without a Path. https://arxiv.science/abs/2611.12103
Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite arXiv Science by Indiciae: BibTeX · RIS.