Neural BRDF Representation and Importance Sampling

TL;DR

This paper introduces a neural network-based BRDF representation that achieves high accuracy and efficient rendering, enabling better compression and importance sampling for realistic material appearance in computer graphics.

Contribution

It proposes a compact neural encoding of BRDFs with adaptive sampling and a novel embedding for importance sampling, improving over prior methods in accuracy and efficiency.

Findings

High-fidelity BRDF reconstruction with neural networks

Efficient importance sampling via learned embeddings

Effective encoding of real-world isotropic and anisotropic BRDFs

Abstract

Controlled capture of real-world material appearance yields tabulated sets of highly realistic reflectance data. In practice, however, its high memory footprint requires compressing into a representation that can be used efficiently in rendering while remaining faithful to the original. Previous works in appearance encoding often prioritised one of these requirements at the expense of the other, by either applying high-fidelity array compression strategies not suited for efficient queries during rendering, or by fitting a compact analytic model that lacks expressiveness. We present a compact neural network-based representation of BRDF data that combines high-accuracy reconstruction with efficient practical rendering via built-in interpolation of reflectance. We encode BRDFs as lightweight networks, and propose a training scheme with adaptive angular sampling, critical for the accurate reconstruction of specular highlights. Additionally, we propose a novel approach to make our representation amenable to importance sampling: rather than inverting the trained networks, we learn to encode them in a more compact embedding that can be mapped to parameters of an analytic BRDF for which importance sampling is known. We evaluate encoding results on isotropic and anisotropic BRDFs from multiple real-world datasets, and importance sampling performance for isotropic BRDFs mapped to two different analytic models.