Research map / Image generation / Editing and style transfer
Style Your Hair: Latent Optimization for Pose-Invariant Hairstyle Transfer via Local-Style-Aware Hair Alignment
Taewoo Kim, Chaeyeon Chung, Yoonseo Kim, Sunghyun Park, Kangyeol Kim, Jaegul Choo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add target hair alignment step (ablation) | Style Your Hair · FFHQ hairstyle transfer | FID: 43.37 -> 39.68 | Improved |
| add semantic label of occluded regions to source segmentation (ablation) | Style Your Hair · FFHQ hairstyle transfer | FID: 39.68 -> 32.69 | Improved |
| add independent source inpainting optimization step (ablation) | Style Your Hair · FFHQ hairstyle transfer | FID: 32.69 -> 18.02 | Improved |
| add local-style-matching loss (ablation) | Style Your Hair · VoxCeleb2 reconstruction | LPIPS: 0.2082 -> 0.2078 | Improved |
| add local-style-matching loss (ablation) | Style Your Hair · VoxCeleb2 reconstruction | SSIM: 0.7716 -> 0.7717 | No clear effect |
| add regularization loss L_reg (ablation) | Style Your Hair · VoxCeleb2 reconstruction | SSIM: 0.7667 -> 0.7716 | Improved |
| add regularization loss L_reg (ablation) | Style Your Hair · VoxCeleb2 reconstruction | LPIPS: 0.2125 -> 0.2082 | Improved |
| Ours vs Barbershop at large pose difference | Style Your Hair · FFHQ Difficult pose difference | FID: 26.30 -> 22.72 | Improved |
| Ours vs Barbershop at small pose difference | Style Your Hair · FFHQ Easy pose difference | FID: 20.75 -> 20.79 | No clear effect |
| Ours vs HairFIT on K-hairstyle reconstruction | Style Your Hair · K-hairstyle | LPIPS: 0.2054 -> 0.1786 | Improved |
| Ours vs HairFIT on VoxCeleb2 reconstruction | Style Your Hair · VoxCeleb2 | LPIPS: 0.2033 -> 0.2078 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- LOHO — LOHO: Latent Optimization of Hairstyles via Orthogonalization (CVPR 2021) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- HairFastGAN: Realistic and Robust Hair Transfer with a Fast Encoder-Based Approach (NeurIPS 2024 · similarity 0.92)
- What to Preserve and What to Transfer: Faithful, Identity-Preserving Diffusion-based Hairstyle Transfer (AAAI 2025 · similarity 0.91)
- Stable-Hair: Real-World Hair Transfer via Diffusion Model (AAAI 2025 · similarity 0.88)
- HairDiffusion: Vivid Multi-Colored Hair Editing via Latent Diffusion (NeurIPS 2024 · similarity 0.88)
- Perm: A Parametric Representation for Multi-Style 3D Hair Modeling (ICLR 2025 · similarity 0.87)
- DiffLocks: Generating 3D Hair from a Single Image using Diffusion Models (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
