Research map / Image generation / Editing and style transfer
ConsistentID: Portrait Generation With Multimodal Fine-Grained Identity Preserving
Jiehui Huang, Xiao ming Dong, Wu Song, Chong Zheng, Zhenchao Tang, Jun Zhou, Yuhao Cheng, Long Chen, Hanhui Li, Yiqiang Yan, Shengcai Liao, Xiaodan Liang
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 facial attention localization loss L_facial (ablation) | ConsistentID · Ablation (Table 2) | DINO: 77.8 -> 86.1 | Improved |
| add facial attention localization loss L_facial (ablation) | ConsistentID · Ablation (Table 2) | CLIP-I: 66.4 -> 75.5 | Improved |
| add facial attention localization loss L_facial (ablation) | ConsistentID · Ablation (Table 2) | FGIS: 82.9 -> 85.6 | Improved |
| add fine-grained features to overall facial features (ablation) | ConsistentID · Ablation (Table 2) | DINO: 80.7 -> 86.1 | Improved |
| fine-grained features alone vs combined with overall facial ID (ablation) | ConsistentID · Ablation (Table 2) | DINO: 86.6 vs 86.1 | No clear effect |
| add LLaVA1.5 descriptions in training (ablation) | ConsistentID · Ablation (Table 2) | CLIP-I: 67.4 -> 75.3 | Improved |
| add LLaVA1.5 descriptions in training (ablation) | ConsistentID · Ablation (Table 2) | DINO: 84.3 -> 85.4 | Improved |
| train image projection jointly with facial encoder (ablation) | ConsistentID · Ablation (Table 2) | DINO: 75.6 -> 86.1 | Improved |
| train image projection jointly with facial encoder (ablation) | ConsistentID · Ablation (Table 2) | CLIP-I: 61.0 -> 75.5 | Improved |
| replace SD1.5 base with SDXL (ablation) | ConsistentID · SDXL extension (180 images, 45 prompts) | FaceSim: 62.9 -> 73.1 | Improved |
| replace SD1.5 base with SDXL (ablation) | ConsistentID · SDXL extension (180 images, 45 prompts) | CLIP-T: 19.8 -> 21.9 | Improved |
| replace SD1.5 base with SDXL (ablation) | ConsistentID · SDXL extension (180 images, 45 prompts) | Speed (s): 16 -> 18 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Photomaker — PhotoMaker: Customizing Realistic Human Photos via Stacked ID Embedding (CVPR 2024) · better on 4 of 4 shared results
- Fastcomposer — FastComposer: Tuning-Free Multi-subject Image Generation with Localized Attention (IJCV 2024) · 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
- FaithfulFaces: Pose-Faithful Facial Identity Preservation for Text-to-Video Generation (NeurIPS 2026 · similarity 0.89)
- FaceA-Net: Facial Attribute-Driven ID Preserving Image Generation Network (AAAI 2025 · similarity 0.89)
- ID-Patch: Robust ID Association for Group Photo Personalization (CVPR 2025 · similarity 0.88)
- RealisID: Scale-Robust and Fine-Controllable Identity Customization via Local and Global Complementation (AAAI 2025 · similarity 0.88)
- HiFi-Portrait: Zero-shot Identity-preserved Portrait Generation with High-fidelity Multi-face Fusion (CVPR 2025 · similarity 0.88)
- When StyleGAN Meets Stable Diffusion: a W+ Adapter for Personalized Image Generation (CVPR 2024 · similarity 0.87)
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.
