LightsOut: Diffusion-based Outpainting for Enhanced Lens Flare Removal

Shr-Ruei Tsai, Wei-Cheng Chang, Jie-Ying Lee, Chih-Hai Su, Yu-Lun Liu
10/17/2025
cs.CV

No official code found. Best implementation: Ray-1026/LightsOut-official (31 stars, MIT, last commit 2025-10-20).

Abstract

Lens flare significantly degrades image quality, impacting critical computer vision tasks like object detection and autonomous driving. Recent Single Image Flare Removal (SIFR) methods perform poorly when off-frame light sources are incomplete or absent. We propose LightsOut, a diffusion-based outpainting framework tailored to enhance SIFR by reconstructing off-frame light sources. Our method leverages a multitask regression module and LoRA fine-tuned diffusion model to ensure realistic and physically consistent outpainting results. Comprehensive experiments demonstrate LightsOut consistently boosts the performance of existing SIFR methods across challenging scenarios without additional retraining, serving as a universally applicable plug-and-play preprocessing solution. Project page: https://ray-1026.github.io/lightsout/

View on arXivView PDF

Code Implementations(1)

Ray-1026/LightsOut-officialCommunity70%
MIT

[ICCV 2025] LightsOut: Diffusion-based Outpainting for Enhanced Lens Flare Removal

312Shell, PythonJul 9, 20251 years agoMIT

Similar Papers

Controllable Light Diffusion for Portraits

83% match2023

Retinex-Diffusion: On Controlling Illumination Conditions in Diffusion Models via Retinex Theory

83% match2024

Be-Your-Outpainter: Mastering Video Outpainting through Input-Specific Adaptation

83% match2 repos2552024

Learning Illumination Control in Diffusion Models

82% match2026

DiLightNet: Fine-grained Lighting Control for Diffusion-based Image Generation

82% match1 repo2092024

ExtraNeRF: Visibility-Aware View Extrapolation of Neural Radiance Fields with Diffusion Models

82% match2024

Cite this paper

@article{tsai2025lightsout,
  title  = {LightsOut: Diffusion-based Outpainting for Enhanced Lens Flare Removal},
  author = {Shr-Ruei Tsai and Wei-Cheng Chang and Jie-Ying Lee and Chih-Hai Su and Yu-Lun Liu},
  year   = {2025},
  eprint = {2510.15868},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2510.15868v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2510.15868.svg)](https://codeofpaper.com/paper/2510.15868)

Discussion