Lens Flare Removal and Reconstruction

2026-09-30

AI Summary

This study addresses the challenge that large lens flares severely disrupt multi-view 3D reconstruction, compounded by the absence of consistent representation and reconstruction methodologies. To tackle this, we construct the first large-scale reflective flare dataset and leverage diffusion models to eliminate complex flare artifacts. Furthermore, this work proposes a symmetry-based explicit representation ensuring multi-view consistency, which integrates a procedural generation pipeline with 3D Gaussian Splatting for joint optimization, thereby achieving decoupled modeling of scenes and flares. Our contributions significantly enhance both flare removal quality and the robustness of scene-flare decomposition, while enabling flexible flare editing and generalization to novel scenes.

Abstract

The presence of lens flares in images can significantly reduce the quality of downstream application results for tasks such as 3D scene reconstruction. This is because lens flares are a property of the camera imaging system, and not a part of the underlying scene being modeled. There are previous methods that tackle the removal of small flares focused around a light source. However, existing methods struggle with large flares, such as those that fill the entire image. In this work, we compile a novel dataset for large-flare removal, combining publicly available real-world data with a procedural generation pipeline. We fine-tune a diffusion-based model on our dataset to remove complex, large lens flares. On the other hand, lens flares remain effective artistic tools, widely used in the media. While there are ways to simulate 2D flares, representing and reconstructing lens flares consistently across multiple views has not yet been explored. To achieve this, we introduce a flare representation model that leverages the symmetry of lens flares about the camera's principal point. We propose a computational pipeline to jointly optimize this flare model and a Gaussian splatting model (3DGS). This enables the decomposition of a 3D scene into lens flares and the scene itself, using our flare-removal model. Because the reconstructed flare is explicit and re-renderable, it can be edited and transferred to novel images and new 3D scenes. We evaluate removal on an established benchmark and a new one for large reflective flares, quantify the flare/scene decomposition directly, and show that the pipeline is robust to errors in automatic light-source localization.

Problem

• lens flare removal
• large lens flare
• flare reconstruction
• multi-view consistency
• scene decomposition

Innovation

• Lens Flare Removal
• Diffusion Model
• Flare Representation
• Gaussian Splatting
• Scene Decomposition