We present a new null-scattering formulation of Smith microfacet BSDFs with unbiased multiple scattering that supports arbitrary NDFs on the full sphere. In contrast to heightfield NDFs that are constrained to the upper hemisphere and become spiky and exhibit a sheen-like quality at high roughnesses (top-right), a full-sphere NDF like a
spherical Gaussian (vMF) transitions smoothly from low-roughness Beckmann into highly porous material. While the two NDFs are very similar at moderate roughnesses (left and middle), note how the vMF BRDF of rough gold transitions smoothly into a diffuse appearance for extreme roughnesses (bottom-right).
Abstract
We present a practical importance-sampling scheme for the Student-T distribution of visible normals by representing the Student-T NDF as a superposition of Beckmann NDFs. Additionally, we derive a new form of delta tracking to evaluate and sample exact BSDFs with general full-sphere NDFs. These tools permit efficient computation of benchmark BSDF values for the multiple scattering from general (including porous) rough surfaces.
Summary
In this talk, we explore the utilization of more generalized Normal Distribution Functions (NDFs) beyond the conventional GGX and Beckmann. The discussion is split into two key parts.
Motivation
We show that previous measurements of light transmittance through rough glass are very well approximated by microfacet BSDFs with the Student-T NDF.
New Algorithms
To incorporate more general NDFs in multiple-scattering microfacet BSDFs we:
Derive analytic sampling for the Student-T NDF (specifically for its vNDF)
Present a new null-scattering algorithm for evaluating and sampling BRDFs with full-sphere NDFs (allowing porous materials)
Demonstrate a new BSDF/NDF interface that allows both biscale roughness and adding new NDFs by implementing one simple function
Citation
@inproceedings{deon2023studentt,
author = {d'Eon, Eugene},
title = {Student-T and Beyond: Practical Tools for Multiple-Scattering BSDFs with General NDFs},
year = {2023},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3587421.3595417},
doi = {10.1145/3587421.3595417},
booktitle = {SIGGRAPH 2023 Talks},
numpages = {2},
location = {Los Angeles, CA, USA}
}
NVIDIA Research
