Learning Diverse Tone Styles for Image Retouching

TL;DR

This paper introduces a novel style-disentangled approach using normalizing flows for diverse image retouching, enabling flexible and subjective aesthetic preferences with stable style representations.

Contribution

It proposes a style domain learning framework with a style encoder, RetouchNet, and TSFlow for disentangling and generating diverse retouching styles, improving over deterministic models.

Findings

Outperforms state-of-the-art methods on MIT-Adobe FiveK and PPR10K datasets.

Effectively generates diverse retouching results matching human aesthetic preferences.

Disentangles style from content for stable and flexible image retouching.

Abstract

Image retouching, aiming to regenerate the visually pleasing renditions of given images, is a subjective task where the users are with different aesthetic sensations. Most existing methods deploy a deterministic model to learn the retouching style from a specific expert, making it less flexible to meet diverse subjective preferences. Besides, the intrinsic diversity of an expert due to the targeted processing on different images is also deficiently described. To circumvent such issues, we propose to learn diverse image retouching with normalizing flow-based architectures. Unlike current flow-based methods which directly generate the output image, we argue that learning in a style domain could (i) disentangle the retouching styles from the image content, (ii) lead to a stable style presentation form, and (iii) avoid the spatial disharmony effects. For obtaining meaningful image tone style representations, a joint-training pipeline is delicately designed, which is composed of a style encoder, a conditional RetouchNet, and the image tone style normalizing flow (TSFlow) module. In particular, the style encoder predicts the target style representation of an input image, which serves as the conditional information in the RetouchNet for retouching, while the TSFlow maps the style representation vector into a Gaussian distribution in the forward pass. After training, the TSFlow can generate diverse image tone style vectors by sampling from the Gaussian distribution. Extensive experiments on MIT-Adobe FiveK and PPR10K datasets show that our proposed method performs favorably against state-of-the-art methods and is effective in generating diverse results to satisfy different human aesthetic preferences. Source code and pre-trained models are publicly available at https://github.com/SSRHeart/TSFlow.