Neural Path Guiding with Distribution Factorization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Figueiredo, Pedro, He, Qihao, Kalantari, Nima Khademi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916780259147776
author Figueiredo, Pedro
He, Qihao
Kalantari, Nima Khademi
author_facet Figueiredo, Pedro
He, Qihao
Kalantari, Nima Khademi
contents In this paper, we present a neural path guiding method to aid with Monte Carlo (MC) integration in rendering. Existing neural methods utilize distribution representations that are either fast or expressive, but not both. We propose a simple, but effective, representation that is sufficiently expressive and reasonably fast. Specifically, we break down the 2D distribution over the directional domain into two 1D probability distribution functions (PDF). We propose to model each 1D PDF using a neural network that estimates the distribution at a set of discrete coordinates. The PDF at an arbitrary location can then be evaluated and sampled through interpolation. To train the network, we maximize the similarity of the learned and target distributions. To reduce the variance of the gradient during optimizations and estimate the normalization factor, we propose to cache the incoming radiance using an additional network. Through extensive experiments, we demonstrate that our approach is better than the existing methods, particularly in challenging scenes with complex light transport.
format Preprint
id arxiv_https___arxiv_org_abs_2506_00839
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Neural Path Guiding with Distribution Factorization
Figueiredo, Pedro
He, Qihao
Kalantari, Nima Khademi
Graphics
Computer Vision and Pattern Recognition
Machine Learning
In this paper, we present a neural path guiding method to aid with Monte Carlo (MC) integration in rendering. Existing neural methods utilize distribution representations that are either fast or expressive, but not both. We propose a simple, but effective, representation that is sufficiently expressive and reasonably fast. Specifically, we break down the 2D distribution over the directional domain into two 1D probability distribution functions (PDF). We propose to model each 1D PDF using a neural network that estimates the distribution at a set of discrete coordinates. The PDF at an arbitrary location can then be evaluated and sampled through interpolation. To train the network, we maximize the similarity of the learned and target distributions. To reduce the variance of the gradient during optimizations and estimate the normalization factor, we propose to cache the incoming radiance using an additional network. Through extensive experiments, we demonstrate that our approach is better than the existing methods, particularly in challenging scenes with complex light transport.
title Neural Path Guiding with Distribution Factorization
topic Graphics
Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2506.00839