Lumos3D: A Single-Forward Framework for Low-Light 3D Scene Restoration

TL;DR

Lumos3D introduces a pose-free, single-forward framework for restoring low-light 3D scenes from multi-view images without scene-specific training, using a novel distillation scheme and Lumos loss.

Contribution

It presents a novel, pose-free, single-forward approach for 3D low-light scene restoration that does not require scene-specific optimization or camera pose estimation.

Findings

Achieves competitive results on real-world datasets.

Operates without per-scene training or optimization.

Uses a cross-illumination distillation scheme and Lumos loss.

Abstract

Restoring 3D scenes with low-light conditions is challenging, and most existing methods depend on precomputed camera poses and scene-specific optimization, which greatly restricts their application to real-world scenarios. To overcome these limitations, we propose Lumos3D, a pose-free single-forward framework for 3D low-light scene restoration. First, we develop a cross-illumination distillation scheme, where a frozen teacher network takes normal-light ground truth images as input to distill accurate geometric information to the student model. Second, we define a Lumos loss to improve the restoration quality of the reconstructed 3D Gaussian space. Trained on a single dataset, Lumos3D performs inference in a purely feed-forward manner, directly restoring illumination and structure from unposed, low-light multi-view images without any per-scene training or optimization. Experiments on real-world datasets demonstrate that Lumos3D achieves competitive restoration results compared to scene-specific methods. Our codes will be released soon.