Research map / Image generation / Text-to-image diffusion
Tuning-Free Image Customization with Image and Text Guidance
Pengzhi Li, Qiang Nie, Ying Chen, Xi Jiang, Kai Wu, Yuhuan Lin, Yong Liu, Jinlong Peng, Chengjie Wang, Feng Zheng
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 blended self-attention to inversion-only baseline (ablation) | tuning-free customization · custom benchmark | DINO: 32.30 -> 50.43 | Improved |
| add enhancement with alpha=1/4, beta=1/2 (ablation) | blended self-attention · custom benchmark | CLIP-T: 26.16 -> 26.86 | Improved |
| use equal weights alpha=beta=1/3 in enhancement (ablation) | blended self-attention enhancement · custom benchmark | DINO: 51.18 -> 47.90 | Hurt |
| apply enhancement over the whole denoising schedule (0, T) (ablation) | blended self-attention enhancement · custom benchmark | DINO: 51.18 -> 39.08 | Hurt |
| apply blended attention over the whole schedule instead of a mid-range window (ablation) | blended self-attention · custom benchmark | DINO: 50.43 -> 47.32 | Hurt |
| replace DDIM inversion with DPM-Solver++ inversion (ablation) | collage inversion · 300 scenes reconstruction | MAE: 0.128 -> 0.041 | Improved |
| use text plus image guidance instead of image-only AnyDoor | image customization · custom benchmark | CLIP-T: 21.22 -> 26.86 | Improved |
| use proposed method instead of two-step compositions and editing baselines | image customization · user study | Fidelity (1-5): 3.91 -> 4.02 | Improved |
Closest papers by meaning
- Towards Understanding Cross and Self-Attention in Stable Diffusion for Text-Guided Image Editing (CVPR 2024 · similarity 0.90)
- Direct Consistency Optimization for Robust Customization of Text-to-Image Diffusion models (NeurIPS 2024 · similarity 0.89)
- Navigating Text-To-Image Customization: From LyCORIS Fine-Tuning to Model Evaluation (ICLR 2024 · similarity 0.88)
- Multi-party Collaborative Attention Control for Image Customization (CVPR 2025 · similarity 0.88)
- How to Continually Adapt Text-to-Image Diffusion Models for Flexible Customization? (NeurIPS 2024 · similarity 0.88)
- PaRa: Personalizing Text-to-Image Diffusion via Parameter Rank Reduction (ICLR 2025 · 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.
