A unified transport model for correlated disorder: microfacets, volumes, porous materials, and beyond

1Dartmouth College 2NVIDIA Research

In International Conference on Transport Theory (ICTT), 2026

Teaser
Abstract Downloads Cite Related

Abstract

In this work, we will describe our recent developments to unify the distinct stochastic appearance models used in computer graphics for surfaces and volumes. By representing scene geometry as the zero crossings of a Gaussian random field, our theory subsumes Beckmann microfacet surfaces, classical participating media, and an exciting new continuum in between, including non-classical media with arbitrary heterogeneity and porous aggregates with correlated microstructures. Each object shown above is represented as a 3D volume that encodes the mean and covariance of a non-stationary Gaussian process. The insets show example realizations of the process. Note how the appearance of the stochastic geometry transitions from volumetric (left) to hard-surface (right) as the correlations in the process are strengthened. Increasing the variance of the process allows us to visualize uncertainty at the macro-scale (bottom).

To efficiently compute light transport in such correlated disorder (without homogenizing to classical models) we interweave the Monte Carlo transport process with “just-in-time” random sampling of 1D slices of the Gaussian process along rays. We also establish a formal connection between Gaussian processes and specific forms of procedural noise commonly used in computer graphics. This allows sampling realizations without expensive linear system solves, all while maintaining precise control over spatial correlations.

We believe our approach, while initially designed for computer graphics applications, holds potential for fields like reactor engineering and remote sensing where dense, correlated, porous media are prevalent.

Downloads

Cite

Wojciech Jarosz, Dario Seyb, Kehan Xu, Eugene d'Eon, Benedikt Bitterli. A unified transport model for correlated disorder: microfacets, volumes, porous materials, and beyond. International Conference on Transport Theory (ICTT), June 2026. 
@inproceedings{jarosz26unified-ictt,
    author    = {Jarosz, Wojciech and Seyb, Dario and Xu, Kehan and d'Eon, Eugene and Bitterli, Benedikt},
    title     = {A unified transport model for correlated disorder: Microfacets, volumes, porous materials, and
                 beyond},
    booktitle = {International Conference on Transport Theory (ICTT)},
    year      = {2026},
    month     = jun,
    location  = {Aix-en-Provence, France}
}