Research map / Image generation / Text-to-image diffusion
Inv-Adapter: ID Customization Generation via Image Inversion and Lightweight Parameter Adapter
Peng Xing, Ning Wang, Jianbo Ouyang, Zechao Li
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 |
|---|---|---|---|
| insert diffusion features into both self and cross attention layers instead of only one (ablation) | Inv-Adapter EAA · face generation | ID fidelity | Improved |
| add token aggregation in the cross attention layer (ablation) | Inv-Adapter · face generation | ID fidelity | Improved |
| add token aggregation in the self attention layer (ablation) | Inv-Adapter · face generation | image quality | Hurt |
| increase aggregation number s_c (ablation) | Inv-Adapter · face generation | FACE-SIM vs CLIP-T | Mixed |
| use Inv-Adapter instead of InstantID | InstantID · Sample-1K | CLIP-I: 70.25 -> 71.81 | Improved |
| use Inv-Adapter instead of InstantID | InstantID · Sample-1K | FACE-SIM: 67.22 -> 65.12 | Hurt |
| use Inv-Adapter instead of IP-Adapter-FaceID-Plus | IP-Adapter-FaceID-Plus · Celebrity | FID: 209.86 -> 201.38 | Improved |
| use Inv-Adapter instead of IP-Adapter-Plus | IP-Adapter-Plus · Sample-1K | FACE-SIM: 37.45 -> 65.12 | Improved |
| use SD v1.5 Inv-Adapter instead of SDXL InstantID | InstantID · Celebrity | FACE-SIM: 69.83 -> 65.42 | Hurt |
| use Inv-Adapter instead of PhotoMaker | PhotoMaker · Sample-1K | FACE-SIM: 26.85 -> 65.12 | Improved |
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.
- PhotoMaker — PhotoMaker: Customizing Realistic Human Photos via Stacked ID Embedding (CVPR 2024) · better on 5 of 5 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
- PhotoMaker: Customizing Realistic Human Photos via Stacked ID Embedding (CVPR 2024 · similarity 0.86)
- CustAny: Customizing Anything from A Single Example (CVPR 2025 · similarity 0.84)
- DisEnvisioner: Disentangled and Enriched Visual Prompt for Customized Image Generation (ICLR 2025 · similarity 0.84)
- PuLID: Pure and Lightning ID Customization via Contrastive Alignment (NeurIPS 2024 · similarity 0.83)
- InstantBooth: Personalized Text-to-Image Generation without Test-Time Finetuning (CVPR 2024 · similarity 0.83)
- Origin Identification for Text-Guided Image-to-Image Diffusion Models (ICML 2025 · similarity 0.83)
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.
