PureSample: Neural Materials Learned by Sampling Microgeometry

2025-08-10

AI Summary

Traditional physics-based BRDF models rely on complex analytical derivations, struggle to represent spatially varying materials, and require model-specific designs for importance sampling and PDF evaluation. To address these limitations, we propose a neural micro-geometry random-walk framework: a flow-matching network jointly learns view-dependent albedo and sampling distributions, enabling end-to-end optimization of material representations; forward random walks generate training data without assuming statistical homogeneity; and a lightweight network unifies shading, importance sampling, and PDF evaluation in a single efficient pass. Our method is validated on multi-layered media, volumetric multiple scattering, and complex micro-structured surfaces. It significantly improves both rendering accuracy and computational efficiency, overcoming the dual constraints of analytical tractability and statistical homogeneity assumptions inherent in conventional approaches.

Technology Category

• Computer Vision: Low Level & Physics-based Vision
• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Learning with Manifolds

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Web Mining and Content Analysis: Web data generation and simulation

Abstract

Traditional physically-based material models rely on analytically derived bidirectional reflectance distribution functions (BRDFs), typically by considering statistics of micro-primitives such as facets, flakes, or spheres, sometimes combined with multi-bounce interactions such as layering and multiple scattering. These derivations are often complex and model-specific, and typically consider a statistical aggregate of a large surface area, ignoring spatial variation. Once an analytic BRDF's evaluation is defined, one still needs to design an importance sampling method for it, and a way to evaluate the pdf of that sampling distribution, requiring further model-specific derivations. We present PureSample: a novel neural BRDF representation that allows learning a material's behavior purely by sampling forward random walks on the microgeometry, which is usually straightforward to implement. Our representation allows for efficient importance sampling, pdf evaluation, and BRDF evaluation, for homogeneous as well as spatially varying materials. We achieve this by two learnable components: first, the sampling distribution is modeled using a flow matching neural network, which allows both importance sampling and pdf evaluation; second, we introduce a view-dependent albedo term, captured by a lightweight neural network, which allows for converting a scalar pdf value to a colored BRDF value for any pair of view and light directions. We demonstrate PureSample on challenging materials, including multi-layered materials, multiple-scattering microfacet materials, and various other microstructures.

Problem

• Develop neural BRDF representation for material behavior learning
• Enable efficient importance sampling and pdf evaluation
• Support homogeneous and spatially varying materials

Innovation

• Neural BRDF learned by sampling microgeometry walks
• Flow matching network for sampling and pdf evaluation
• View-dependent albedo term for colored BRDF conversion