Research map / Image generation / Text-to-image diffusion
Mod-Adapter: Tuning-Free and Versatile Multi-concept Personalization via Modulation Adapter
Weizhi Zhong, Huan Yang, Zheng Liu, Huiguo He, Zijian He, Xuesong Niu, Di ZHANG, Guanbin 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 |
|---|---|---|---|
| add VLM-guided pre-training (ablation) | Mod-Adapter · DreamBench-Abs (multi-concept) | CP·PF: 0.17 -> 0.62 | Improved |
| add VLM-guided pre-training (ablation) | Mod-Adapter · DreamBench-Abs (single-concept) | CP·PF: 0.24 -> 0.54 | Improved |
| add Mixture-of-Experts layer (ablation) | Mod-Adapter · DreamBench-Abs (multi-concept) | CP·PF: 0.35 -> 0.62 | Improved |
| add Mixture-of-Experts layer (ablation) | Mod-Adapter · DreamBench-Abs (single-concept) | CP·PF: 0.42 -> 0.54 | Improved |
| add VL cross-attention (ablation) | Mod-Adapter · DreamBench-Abs (multi-concept) | CP·PF: 0.39 -> 0.62 | Improved |
| add VL cross-attention (ablation) | Mod-Adapter · DreamBench-Abs (single-concept) | CP·PF: 0.49 -> 0.54 | Improved |
| add VL cross-attention (ablation) | Mod-Adapter · DreamBench-Abs (multi-concept) | CLIP-T: 0.329 -> 0.330 | No clear effect |
| use k-means routing instead of learnable linear gating (ablation) | Mod-Adapter · DreamBench-Abs (multi-concept) | CP·PF: 0.49 -> 0.62 | Improved |
| use k-means routing instead of learnable linear gating (ablation) | Mod-Adapter · DreamBench-Abs (single-concept) | CP·PF: 0.44 -> 0.54 | Improved |
| increase number of experts from 1 to 12 (ablation) | Mod-Adapter · DreamBench-Abs (multi-concept) | CP·PF: 0.22 -> 0.62 | Improved |
| increase number of experts from 4 to 12 (ablation) | Mod-Adapter · DreamBench-Abs (single-concept) | CP·PF: 0.31 -> 0.54 | Improved |
| use hierarchical instead of k-means clustering (ablation) | Mod-Adapter · DreamBench-Abs (multi-concept) | CP·PF: 0.62 -> 0.63 | No clear effect |
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.
- MS-Diffusion — MS-Diffusion: Multi-subject Zero-shot Image Personalization with Layout Guidance (ICLR 2025) · 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
- Concept Weaver: Enabling Multi-Concept Fusion in Text-to-Image Models (CVPR 2024 · similarity 0.90)
- MC^2: Multi-concept Guidance for Customized Multi-concept Generation (CVPR 2025 · similarity 0.89)
- How to Continually Adapt Text-to-Image Diffusion Models for Flexible Customization? (NeurIPS 2024 · similarity 0.88)
- Concept Conductor: Orchestrating Multiple Personalized Concepts in Text-to-Image Synthesis (AAAI 2025 · similarity 0.88)
- Yo'Chameleon: Personalized Vision and Language Generation (CVPR 2025 · similarity 0.88)
- ClassDiffusion: More Aligned Personalization Tuning with Explicit Class Guidance (ICLR 2025 · similarity 0.88)
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.
