FlareX: A Physics-Informed Dataset for Lens Flare Removal via 2D Synthesis and 3D Rendering

2025-10-10

AI Summary

Lens flare removal suffers from scarce real paired data and poor generalization due to physically inaccurate and insufficiently diverse synthetic data. To address this, we propose a hybrid data generation paradigm grounded in illumination physics: (1) parameterized design coupled with illumination-aware mechanisms generates highly diverse 2D flare templates; (2) physically-based 3D scene rendering is performed, integrated with a mask-guided strategy to extract authentic flare-free images. This yields FlareX—a high-quality dataset comprising 9,500 2D templates and 3,000 paired 3D-rendered images. Experiments demonstrate that FlareX substantially improves the generalization of flare removal models on real-world scenes. Notably, our work establishes the first physics-driven, joint 2D/3D flare modeling and evaluation framework, bridging the gap between synthetic fidelity and real-world applicability.

Technology Category

• Computer Vision: 3D Computer Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Lens flare occurs when shooting towards strong light sources, significantly degrading the visual quality of images. Due to the difficulty in capturing flare-corrupted and flare-free image pairs in the real world, existing datasets are typically synthesized in 2D by overlaying artificial flare templates onto background images. However, the lack of flare diversity in templates and the neglect of physical principles in the synthesis process hinder models trained on these datasets from generalizing well to real-world scenarios. To address these challenges, we propose a new physics-informed method for flare data generation, which consists of three stages: parameterized template creation, the laws of illumination-aware 2D synthesis, and physical engine-based 3D rendering, which finally gives us a mixed flare dataset that incorporates both 2D and 3D perspectives, namely FlareX. This dataset offers 9,500 2D templates derived from 95 flare patterns and 3,000 flare image pairs rendered from 60 3D scenes. Furthermore, we design a masking approach to obtain real-world flare-free images from their corrupted counterparts to measure the performance of the model on real-world images. Extensive experiments demonstrate the effectiveness of our method and dataset.

Problem

• Existing datasets lack flare diversity and physical accuracy
• Current methods fail to generalize to real-world scenarios
• Need physics-informed approach for realistic flare removal data

Innovation

• Physics-informed flare generation via 2D synthesis
• 3D rendering using physical engine for flare diversity
• Masking approach extracts flare-free images from corrupted ones