Neural Visibility Cache for Real-Time Light Sampling

TL;DR

This paper introduces a neural cache that efficiently stores light visibility information for real-time physically-based rendering, improving light sampling performance in complex scenes.

Contribution

It presents a novel online-trained neural cache using a multi-resolution hash-grid encoding for real-time light visibility storage and sampling.

Findings

Achieves real-time inference on modern GPUs

Integrates seamlessly with existing rendering frameworks

Enhances light sampling efficiency in complex scenes

Abstract

Direct illumination with many lights is an inherent component of physically-based rendering, remaining challenging, especially in real-time scenarios. We propose an online-trained neural cache that stores visibility between lights and 3D positions. We feed light visibility to weighted reservoir sampling (WRS) to sample a light source. The cache is implemented as a fully-fused multilayer perceptron (MLP) with multi-resolution hash-grid encoding, enabling online training and efficient inference on modern GPUs in real-time frame rates. The cache can be seamlessly integrated into existing rendering frameworks and can be used in combination with other real-time techniques such as spatiotemporal reservoir sampling (ReSTIR).