Style your hair

Abstract

Editing hairstyle is unique and challenging due to the complexity and delicacy of hairstyle. Although recent approaches significantly improved the hair details, this is achieved under the assumption that a target hair and a source image are aligned. HairFIT, a pose-invariant hairstyle transfer model, alleviates this assumption, yet it still shows unsatisfactory quality in preserving delicate hair textures. To solve these limitations, we propose a high-performing pose-invariant hairstyle transfer model equipped with a latent optimization and a newly presented local-style-matching loss. In the StyleGAN2 latent space, we first explore a pose-aligned latent code of a target hair with the detailed textures preserved based on local-style-matching. Then, our model inpaints the occlusions of the source considering the aligned target hair and blends both images to produce a final output. The experimental results demonstrate that our model has strengths in transferring a hairstyle under higher pose differences and preserving local hairstyle textures.