Research map / Image generation / Editing and style transfer
A Task is Worth One Word: Learning with Task Prompts for High-Quality Versatile Image Inpainting
Junhao Zhuang, Yanhong Zeng, WENRAN LIU, Chun Yuan, Kai Chen
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 |
|---|---|---|---|
| use learnable task prompts instead of unlearnable rare identifiers (ablation) | PowerPaint · Places2 context inpainting | FID: 3.11 -> 2.59 | Improved |
| use learnable task prompts instead of unlearnable rare identifiers (ablation) | PowerPaint · MSCOCO object inpainting (bbox) | Local-FID: 13.03 -> 11.61 | Improved |
| train one unified model with task prompts instead of separate task-specific models (ablation) | PowerPaint · Flickr-Scenery outpainting | FID: 10.24 -> 10.16 | Improved |
| unified model vs task-specific on text-guided object inpainting (ablation) | PowerPaint · OpenImages bbox object inpainting | CLIP Score: 27.58 -> 27.56 | No clear effect |
| fine-tune SD-Inpainting on the same fine-tuning data as PowerPaint (ablation) | SD-Inpainting · OpenImages bbox object inpainting | Local-FID: 12.71 -> 12.83 | No clear effect |
| use object-removal prompt as positive and object-insertion prompt as negative prompt with classifier-free guidance | PowerPaint · context-aware inpainting and object removal | artifact generation: Places2 FID 2.82 -> 2.59 vs SD-Inpainting with default prompt | Improved |
| use PowerPaint instead of SmartBrush for context-aware inpainting and outpainting | SmartBrush · Flickr-Scenery outpainting | FID: 105.99 -> 10.16 | Improved |
| use PowerPaint instead of LaMa for context-aware inpainting | LaMa · Places2 context-aware inpainting (all samples) | LPIPS: 0.1193 -> 0.1263 | Hurt |
| use PowerPaint instead of SmartBrush for text-guided object inpainting | SmartBrush · OpenImages bbox object inpainting | CLIP Score: 27.61 -> 27.56 | 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.
- SmartBrush — SmartBrush: Text and Shape Guided Object Inpainting With Diffusion Model (CVPR 2023) · better on 6 of 7 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
- Brush2Prompt: Contextual Prompt Generator for Object Inpainting (CVPR 2024 · similarity 0.91)
- Paint by Inpaint: Learning to Add Image Objects by Removing Them First (CVPR 2025 · similarity 0.87)
- HD-Painter: High-Resolution and Prompt-Faithful Text-Guided Image Inpainting with Diffusion Models (ICLR 2025 · similarity 0.87)
- TurboFill: Adapting Few-step Text-to-image Model for Fast Image Inpainting (CVPR 2025 · similarity 0.85)
- MVInpainter: Learning Multi-View Consistent Inpainting to Bridge 2D and 3D Editing (NeurIPS 2024 · similarity 0.85)
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency (CVPR 2025 · similarity 0.85)
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.
