When Extreme Darkness Meets Motion Blur: MeanFlow for Unified RAW Restoration
2026-08-03 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors focus on improving photos taken in extremely dark conditions using RAW camera data, which is hard due to both noise and motion blur. They created a new dataset called SIDED that adds motion blur to low-light images to better mimic real-world problems. Their method uses a special RAW tokenizer and a MeanFlow model to enhance the images in one step. They also add a physics-based refinement to keep lighting and colors consistent without slowing down the process. Experiments show their approach works well in fixing both noise and motion blur in very dark photos.
RAW image datalow-light imagingmotion blurimage enhancementsensor noisedatasettokenizerMeanFlowphysics-guided refinementillumination consistency
Authors
Zepu Wang, Jingze Liang, Weijie Xiao, Kexin Chen
Abstract
Extremely low-light RAW enhancement aims to recover severely attenuated sensor signals, yet existing methods often focus on illumination and noise while overlooking the motion-induced degradations inherent in practical low-light imaging. We present a framework for robust extremely low-light RAW enhancement under realistic acquisition degradations. First, we introduce See in the Degraded Extremely Dark (SIDED), a new dataset that applies controlled motion degradation to extremely low-light RAW pairs while retaining their original sensor noise. Second, we propose a unified RAW tokenizer equipped with explicit domain-conditioned representation calibration to align extremely low-light and well-exposed RAW data, followed by a MeanFlow that performs enhancement in a single function evaluation. To our knowledge, this is the first work to formulate extremely low-light RAW enhancement under realistic motion-degraded acquisition and address it with MeanFlow. We further introduce a physics-guided refinement model to strengthen illumination--reflectance consistency, pixel fidelity, and color preservation without incurring additional inference cost. Extensive experiments demonstrate that our framework achieves state-of-the-art performance in extremely low-light RAW enhancement, and robustly handles coupled motion and noise degradations.