CDST: Color Disentangled Style Transfer for Universal Style Reference Customization

TL;DR

CDST introduces a color-disentangled, tuning-free style transfer method that enhances style similarity and preserves content features, enabling universal style transfer with strong editing capabilities.

Contribution

It proposes a novel two-stream training paradigm that isolates color from style, allowing universal, tuning-free style transfer with improved quality and flexibility.

Findings

Achieves state-of-the-art results on style transfer tasks.

Significantly improves style similarity through multi-feature embedding compression.

Preserves content features while enabling strong editing capabilities.

Abstract

We introduce Color Disentangled Style Transfer (CDST), a novel and efficient two-stream style transfer training paradigm which completely isolates color from style and forces the style stream to be color-blinded. With one same model, CDST unlocks universal style transfer capabilities in a tuning-free manner during inference. Especially, the characteristics-preserved style transfer with style and content references is solved in the tuning-free way for the first time. CDST significantly improves the style similarity by multi-feature image embeddings compression and preserves strong editing capability via our new CDST style definition inspired by Diffusion UNet disentanglement law. By conducting thorough qualitative and quantitative experiments and human evaluations, we demonstrate that CDST achieves state-of-the-art results on various style transfer tasks.