Research map / Image generation / Text-to-image diffusion
MasaCtrl: Tuning-Free Mutual Self-Attention Control for Consistent Image Synthesis and Editing
Mingdeng Cao, Xintao Wang, Zhongang Qi, Ying Shan, Xiaohu Qie, Yinqiang 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 |
|---|---|---|---|
| apply mutual self-attention at all denoising steps and all U-Net layers (ablation) | MasaCtrl (Stable Diffusion v1.4) · synthetic and real image editing (qualitative) | layout compliance with target prompt | Hurt |
| start mutual self-attention only after several denoising steps (S=4) instead of from the first step (ablation) | MasaCtrl · qualitative step ablation | layout from target prompt plus content from source | Improved |
| restrict mutual self-attention control to decoder layers (L=10) instead of low-resolution or encoder layers (ablation) | MasaCtrl · qualitative layer ablation | target layout and source content preservation | Improved |
| start control too late in denoising (ablation) | MasaCtrl · qualitative step ablation | source content retention | Hurt |
| add cross-attention derived foreground mask to mutual self-attention (ablation) | MasaCtrl · images with similar foreground and background (qualitative) | foreground/background confusion | Improved |
| integrate MasaCtrl with T2I-Adapter and use earlier start (S=2, L=8) | MasaCtrl · synthesis and real image editing (qualitative) | faithfulness of content and layout | Improved |
| use MasaCtrl instead of P2P, SDEdit or PnP for non-rigid edits | Stable Diffusion editing · synthetic and real images (qualitative) | compliance with target prompt and content consistency | Improved |
Closest papers by meaning
- Towards Understanding Cross and Self-Attention in Stable Diffusion for Text-Guided Image Editing (CVPR 2024 · similarity 0.88)
- Devil is in the Detail: Towards Injecting Fine Details of Image Prompt in Image Generation via Conflict-free Guidance and Stratified Attention (CVPR 2025 · similarity 0.88)
- TextCtrl: Diffusion-based Scene Text Editing with Prior Guidance Control (NeurIPS 2024 · similarity 0.87)
- Multi-party Collaborative Attention Control for Image Customization (CVPR 2025 · similarity 0.87)
- Ctrl-X: Controlling Structure and Appearance for Text-To-Image Generation Without Guidance (NeurIPS 2024 · similarity 0.87)
- Click2Mask: Local Editing with Dynamic Mask Generation (AAAI 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.
