Many-Worlds Inverse Rendering

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Ziyi, Roussel, Nicolas, Jakob, Wenzel
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912351799738368
author Zhang, Ziyi
Roussel, Nicolas
Jakob, Wenzel
author_facet Zhang, Ziyi
Roussel, Nicolas
Jakob, Wenzel
contents Discontinuous visibility changes remain a major bottleneck when optimizing surfaces within a physically-based inverse renderer. Many previous works have proposed sophisticated algorithms and data structures to sample visibility silhouettes more efficiently. Our work presents another solution: instead of differentiating a tentative surface locally, we differentiate a volumetric perturbation of a surface. We refer this as a many-worlds representation because it models a non-interacting superposition of conflicting explanations (worlds) of the input dataset. Each world is optically isolated from others, leading to a new transport law that distinguishes our method from prior work based on exponential random media. The resulting Monte Carlo algorithm is simpler and more efficient than prior methods. We demonstrate that our method promotes rapid convergence, both in terms of the total iteration count and the cost per iteration.
format Preprint
id arxiv_https___arxiv_org_abs_2408_16005
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Many-Worlds Inverse Rendering
Zhang, Ziyi
Roussel, Nicolas
Jakob, Wenzel
Computer Vision and Pattern Recognition
Graphics
Discontinuous visibility changes remain a major bottleneck when optimizing surfaces within a physically-based inverse renderer. Many previous works have proposed sophisticated algorithms and data structures to sample visibility silhouettes more efficiently. Our work presents another solution: instead of differentiating a tentative surface locally, we differentiate a volumetric perturbation of a surface. We refer this as a many-worlds representation because it models a non-interacting superposition of conflicting explanations (worlds) of the input dataset. Each world is optically isolated from others, leading to a new transport law that distinguishes our method from prior work based on exponential random media. The resulting Monte Carlo algorithm is simpler and more efficient than prior methods. We demonstrate that our method promotes rapid convergence, both in terms of the total iteration count and the cost per iteration.
title Many-Worlds Inverse Rendering
topic Computer Vision and Pattern Recognition
Graphics
url https://arxiv.org/abs/2408.16005