Diffusion Restore: Real-Time Markov Chain Monte Carlo Light Transport

2026-05-09

AI Summary

Traditional MCMC light transport methods often suffer from mode stagnation or excessive backtracking during local exploration, struggling to balance sampling efficiency with the discovery of global illumination distributions. This work proposes the Diffusion Restore framework, which for the first time integrates an uncorrected Metropolis-free diffusion process with irreversible Markov chain dynamics, augmenting the Restore approach with a momentum-driven directional exploration mechanism. The method substantially improves sampling efficiency and convergence speed for high-dimensional lighting distributions and is efficiently implemented using GPU ray tracing and compute shaders. Experiments demonstrate that it outperforms existing MCMC techniques across a range of complex scenes, achieves real-time frame rates, and surpasses conventional path tracing in both offline and real-time rendering settings.

Abstract

We present Diffusion Restore, a real-time framework for diffusion-based MCMC light transport. MCMC methods are highly suitable for sampling from complex high-dimensional distributions and for approximating integrals over them. In practice, they are often the only viable solution when direct sampling is not possible and alternative methods are either inefficient or cannot be applied due to the structure of the target distribution. However, controlling the exploration of the target distribution in MCMC methods remains challenging. Efficient exploration requires a balance between local exploration and global discovery, and local dynamics must rapidly explore individual modes without getting stuck or exhibiting excessive backtracking. The problem of global discovery has recently been addressed by the introduction of the Restore framework. In this work, we build on this framework and focus on improving local exploration. We show how to choose diffusion-based local dynamics within the Restore framework while completely avoiding Metropolis-adjustment, which is known to slow down convergence. Furthermore, we model these dynamics as nonreversible, introducing momentum in the drift and thereby enabling more directed exploration of the target distribution compared to reversible, random-walk-like dynamics. We provide a theoretical justification for the validity of our choice of local dynamics. Empirically, we demonstrate across diverse scenes that Diffusion Restore outperforms all existing MCMC light transport methods and establishes a new state of the art. In addition, we present a GPU implementation in ray tracing and compute shaders and achieve real-time frame rates. This demonstrates that Diffusion Restore is not only superior in offline rendering, but also outperforms traditional Path Tracing methods in real-time rendering settings, such as interactive applications and games.

Problem

• MCMC light transport
• local exploration
• real-time rendering
• diffusion dynamics
• nonreversible sampling

Innovation

• Diffusion-based MCMC
• Nonreversible dynamics
• Light transport
• Real-time rendering
• Restore framework