Research map / Image generation / Text-to-image diffusion
MagicMakeup: A Region-Controllable Diffusion Transformer for High-Fidelity Makeup-Transfer
siming Zheng, Siming Zheng, yang yang, Shusong Xu, Hao Zhang, Bo Li, Changqing Zou, Peng-Tao Jiang
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 Token-Aligned Region Gating (TARG) (ablation) | base DiT · MakeupHQ-Synthetic (Eyes) | Face-ID: 0.870 -> 0.890 | Improved |
| add Cross-Modal Perception Guidance (CMPG) on top of TARG (ablation) | MagicMakeup · MakeupHQ-Synthetic (Lip) | Face-ID: 0.885 -> 0.971 | Improved |
| add CMPG on top of TARG (ablation) | MagicMakeup · MakeupHQ-Synthetic (Eyes) | DINO-I: 0.799 -> 0.801 | Improved |
| add TARG (ablation) | base DiT · MakeupHQ-Synthetic (Face) | CLIP-I: 0.833 -> 0.811 | Hurt |
| add TARG to lip region (ablation) | base DiT · MakeupHQ-Synthetic (Lip) | CLIP-I: 0.877 -> 0.870 | Hurt |
| MagicMakeup vs baselines | MagicMakeup · MakeupHQ-Synthetic | L2-M: 111.476 -> 56.373 (vs Flux-Makeup) | Improved |
| MagicMakeup vs Flux-Makeup | MagicMakeup · MakeupHQ-Real | Face-ID: 0.792 -> 0.706 | Hurt |
| MagicMakeup vs Stable-Makeup | MagicMakeup · Makeup-Wild (256x256) | FID: 85.132 -> 92.924 | Hurt |
| perturb masks (translation, rotation, blur) (ablation) | MagicMakeup · masks robustness | Face-ID: 0.979 -> 0.964 | No clear effect |
| user study vs best baseline | MagicMakeup · MakeupHQ user study | selection rate (%): 9.17 -> 77.43 | Improved |
Closest papers by meaning
- Diffusion-Based Makeup Transfer with Facial Region-Aware Makeup Features (CVPR 2026 · similarity 0.91)
- Anchoring on Reality: Breaking the Pseudo-Target Ceiling in Makeup Transfer (ECCV 2026 · similarity 0.90)
- MMFace-DiT: A Dual-Stream Diffusion Transformer for High-Fidelity Multimodal Face Generation (CVPR 2026 · similarity 0.89)
- When StyleGAN Meets Stable Diffusion: a W+ Adapter for Personalized Image Generation (CVPR 2024 · similarity 0.87)
- SHMT: Self-supervised Hierarchical Makeup Transfer via Latent Diffusion Models (NeurIPS 2024 · similarity 0.86)
- DreamSalon: A Staged Diffusion Framework for Preserving Identity-Context in Editable Face Generation (CVPR 2024 · 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.
