Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

VeraRetouch: A Lightweight Fully Differentiable Framework for Multi-Task Reasoning Photo Retouching

About

Reasoning photo retouching has gained significant traction, requiring models to analyze image defects, give reasoning processes, and execute precise retouching enhancements. However, existing approaches often rely on non-differentiable external software, creating optimization barriers and suffering from high parameter redundancy and limited generalization. To address these challenges, we propose VeraRetouch, a lightweight and fully differentiable framework for multi-task photo retouching. We employ a 0.5B Vision-Language Model (VLM) as the central intelligence to formulate retouching plans based on instructions and scene semantics. Furthermore, we develop a fully differentiable Retouch Renderer that replaces external tools, enabling direct end-to-end pixel-level training through decoupled control latents for lighting, global color, and specific color adjustments. To overcome data scarcity, we introduce AetherRetouch-1M+, the first million-scale dataset for professional retouching, constructed via a new inverse degradation workflow. Furthermore, we propose DAPO-AE, a reinforcement learning post-training strategy that enhances autonomous aesthetic cognition. Extensive experiments demonstrate that VeraRetouch achieves state-of-the-art performance across multiple benchmarks while maintaining a significantly smaller footprint, enabling mobile deployment. Our code and models are publicly available at https://github.com/OpenVeraTeam/VeraRetouch.

Yihong Guo, Youwei Lyu, Jiajun Tang, Yizhuo Zhou, Hongliang Wang, Jinwei Chen, Changqing Zou, Qingnan Fan• 2026

Related benchmarks

TaskDatasetResultRank
Photo RetouchingPPR10K-Bench
PSNR24.43
10
Image RetouchingFiveK-Bench
PSNR26.85
8
Auto-RetouchAether-Bench Auto 1M+ (test)
Hist-M89.59
8
Photo RetouchingAether-bench Auto-Syn
Hist-L89.22
8
Automatic Image RetouchingAether-Bench Auto (test)
Total Time (s)6.9
7
Style-RetouchAether-Bench Style 1M+ (test)
L1 Loss0.092
6
Param-RetouchAether-Bench Param 1M+ (test)
L1 Loss0.023
4
Parametric RetouchingAether-Bench Param (test)
Total Time (s)5.17
1
Style RetouchingAether-Bench Style (test)
Total Time (s)3.83
1
Showing 9 of 9 rows

Other info

Follow for update