Manifold Sampling for Differentiable Uncertainty in Radiance Fields

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lyu, Linjie, Tewari, Ayush, Habermann, Marc, Saito, Shunsuke, Zollhöfer, Michael, Leimkühler, Thomas, Theobalt, Christian
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916401030103040
author Lyu, Linjie
Tewari, Ayush
Habermann, Marc
Saito, Shunsuke
Zollhöfer, Michael
Leimkühler, Thomas
Theobalt, Christian
author_facet Lyu, Linjie
Tewari, Ayush
Habermann, Marc
Saito, Shunsuke
Zollhöfer, Michael
Leimkühler, Thomas
Theobalt, Christian
contents Radiance fields are powerful and, hence, popular models for representing the appearance of complex scenes. Yet, constructing them based on image observations gives rise to ambiguities and uncertainties. We propose a versatile approach for learning Gaussian radiance fields with explicit and fine-grained uncertainty estimates that impose only little additional cost compared to uncertainty-agnostic training. Our key observation is that uncertainties can be modeled as a low-dimensional manifold in the space of radiance field parameters that is highly amenable to Monte Carlo sampling. Importantly, our uncertainties are differentiable and, thus, allow for gradient-based optimization of subsequent captures that optimally reduce ambiguities. We demonstrate state-of-the-art performance on next-best-view planning tasks, including high-dimensional illumination planning for optimal radiance field relighting quality.
format Preprint
id arxiv_https___arxiv_org_abs_2409_12661
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Manifold Sampling for Differentiable Uncertainty in Radiance Fields
Lyu, Linjie
Tewari, Ayush
Habermann, Marc
Saito, Shunsuke
Zollhöfer, Michael
Leimkühler, Thomas
Theobalt, Christian
Computer Vision and Pattern Recognition
Graphics
Radiance fields are powerful and, hence, popular models for representing the appearance of complex scenes. Yet, constructing them based on image observations gives rise to ambiguities and uncertainties. We propose a versatile approach for learning Gaussian radiance fields with explicit and fine-grained uncertainty estimates that impose only little additional cost compared to uncertainty-agnostic training. Our key observation is that uncertainties can be modeled as a low-dimensional manifold in the space of radiance field parameters that is highly amenable to Monte Carlo sampling. Importantly, our uncertainties are differentiable and, thus, allow for gradient-based optimization of subsequent captures that optimally reduce ambiguities. We demonstrate state-of-the-art performance on next-best-view planning tasks, including high-dimensional illumination planning for optimal radiance field relighting quality.
title Manifold Sampling for Differentiable Uncertainty in Radiance Fields
topic Computer Vision and Pattern Recognition
Graphics
url https://arxiv.org/abs/2409.12661