Research map / Image generation / Text-to-image diffusion
MagicEraser: Erasing Any Objects via Semantics-Aware Control
Fan Li, Zixiao Zhang, Yi Huang, Jianzhuang Liu, Renjing Pei, Bin Shao, Songcen Xu
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 |
|---|---|---|---|
| train on OLRD shifted-object data instead of random masks (ablation) | SD Inpainting baseline · RealHM | PSNR: 21.331dB -> 22.130dB | Improved |
| add content initialization with LaMa (ablation) | SD Inpainting + OLRD · RealHM | PSNR: 22.130dB -> 22.891dB | Improved |
| add semantics-aware attention refocus (ablation) | OLRD + Content Initialization · RealHM | PSNR: 22.891dB -> 23.277dB | Improved |
| add semantics-aware attention refocus (ablation) | OLRD + Content Initialization · RealHM | FID: 48.91 -> 48.93 | No clear effect |
| add prompt tuning (textual inversion + LoRA) (ablation) | OLRD + Content Initialization · RealHM | PSNR: 22.891dB -> 23.311dB | Improved |
| add both prompt tuning and attention refocus (ablation) | OLRD + Content Initialization · RealHM | FID: 48.91 -> 46.56 | Improved |
| MagicEraser vs SD Inpainting on FID | MagicEraser · RealHM | FID: 45.05 vs 46.56 | Hurt |
| MagicEraser vs MAT on SSIM/LPIPS | MagicEraser · OpenImages | SSIM: 0.949 vs 0.947 | Hurt |
| MagicEraser vs commercial Photoshop Generative Fill | MagicEraser · RealHM | PSNR: 22.913dB -> 23.620dB | Improved |
Closest papers by meaning
- MagicPaint: Operate Anything for Image Inpainting with Diffusion Model (AAAI 2026 · similarity 0.91)
- Inst-Inpaint: Instructing to Remove Objects with Diffusion Models (IJCV 2026 · similarity 0.88)
- CLIPAway: Harmonizing focused embeddings for removing objects via diffusion models (NeurIPS 2024 · similarity 0.88)
- Attentive Eraser: Unleashing Diffusion Model’s Object Removal Potential via Self-Attention Redirection Guidance (AAAI 2025 · similarity 0.88)
- Paint by Inpaint: Learning to Add Image Objects by Removing Them First (CVPR 2025 · similarity 0.86)
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency (CVPR 2025 · similarity 0.86)
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.
