The Devil Is in the Leakage: A Disentangled Dual-Purification Framework for High-Fidelity Hairstyle Transfer

2026-07-13

AI Summary

Existing hair transfer methods struggle in zero-shot settings to disentangle hairstyle from identity and pose information in reference images, often resulting in identity leakage and artifacts in bald regions that compromise synthesis fidelity and identity consistency. To address this, this work proposes a Dual Purification Framework (DPF) that employs a two-stage regularization mechanism: Adversarial Hairstyle Purification (AHP) and Contrastive Geometry Purification (CGP), which respectively purify the hairstyle representation and geometric guidance pathway. DPF is the first approach to systematically identify and resolve the dual leakage problem in hairstyle transfer by integrating diffusion models, ControlNet, mutual information minimization, and contrastive learning to achieve more thorough feature disentanglement. Experiments demonstrate that DPF significantly enhances both photorealism and source identity preservation across multiple benchmarks, achieving state-of-the-art performance.

Abstract

Hairstyle transfer aims to synthesize a photorealistic portrait by transplanting the hairstyle from a reference image onto a source subject while preserving the source identity. Recent foundation models show strong generative capability, but they struggle with the zero-shot disentanglement required for precise local editing, often entangling the reference hairstyle with its original identity and pose. Existing diffusion-based pipelines typically decompose the task by first generating a "bald" image from the source and then injecting hairstyle features from the reference. However, we show that this paradigm suffers from a fundamental leakage problem. Identity Leakage in Hairstyle occurs when hairstyle features retain reference identity or pose information, while Flaw Leakage in Bald arises when residual artifacts in the bald image are propagated into the final synthesis. To address both issues, we propose the Dual-Purification Framework (DPF), which introduces two complementary training-time regularizers. Adversarial Hairstyle Purification (AHP) purifies hairstyle features by suppressing identity predictability under a mutual-information-inspired adversarial objective. Contrastive Geometric Purification (CGP) regularizes the ControlNet pathway with a contrastive objective, reducing the model's reliance on geometric artifacts in the bald condition. By jointly purifying the hairstyle representation and geometric pathway, DPF achieves high-fidelity, identity-preserving hairstyle transfer and state-of-the-art performance on diverse benchmarks.

Problem

• Hairstyle Transfer
• Identity Leakage
• Artifact Leakage
• Disentanglement
• Diffusion Models

Innovation

• Dual-Purification Framework
• Adversarial Hairstyle Purification
• Contrastive Geometric Purification
• Identity Leakage
• Hairstyle Transfer