Research map / Image generation / Editing and style transfer
Inpaint4Drag: Repurposing Inpainting Models for Drag-Based Image Editing via Bidirectional Warping
Jingyi Lu, Kai Han
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 Inpaint4Drag pixel-space warping plus inpainting instead of FastDrag | drag editing · DragBench-S | MD (x100): 4.1 -> 3.6 | Improved |
| use Inpaint4Drag instead of DragDiffusion | drag editing · DragBench-D | MD (x100): 6.7 -> 3.9 | Improved |
| use Inpaint4Drag instead of DragDiffusion for speed | drag editing · 512x512 inference | time (s): 177.7 -> 0.3 | Improved |
| use Inpaint4Drag instead of FastDrag for speed | drag editing · 512x512 inference | time (s): 4.2 -> 0.3 | Improved |
| use Inpaint4Drag instead of DragDiffusion for memory | drag editing · 512x512 inference | GPU memory (GB): 11.6 -> 2.7 | Improved |
| use Inpaint4Drag instead of SDE-Drag for identity preservation | drag editing · DragBench-S | LPIPS (x100): 11.4 -> 11.4 | No clear effect |
| use Inpaint4Drag instead of FastDrag | drag editing · DragBench-S | LPIPS (x100): 24.1 -> 11.4 | Improved |
| replace SD-1.5-Inpaint with DeepFillv2 as inpainting backend (ablation) | Inpaint4Drag · DragBench-S | MD (x100): 3.6 -> 3.2 | Improved |
| replace SD-1.5-Inpaint with LaMa as inpainting backend (ablation) | Inpaint4Drag · DragBench-S | LPIPS (x100): 11.4 -> 13.6 | Hurt |
| replace SD-1.5-Inpaint with SD-XL-Inpaint as inpainting backend (ablation) | Inpaint4Drag · DragBench-D | LPIPS (x100): 9.1 -> 8.8 | Improved |
| replace SD-1.5-Inpaint with DeepFillv2 for efficiency (ablation) | Inpaint4Drag · 512x512 inference | time (s): 0.3 -> 0.05 | Improved |
| replace SD-1.5-Inpaint with SD-XL-Inpaint (ablation) | Inpaint4Drag · 512x512 inference | time (s): 0.3 -> 1.3 | Hurt |
Later papers that compared with this method
2 later accepted papers report Inpaint4Drag as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- In-context Region-based Drag: Drag Any Region to Any Shape (ECCV 2026 · better on 5 of 5 shared results)
- LazyDrag: Enabling Stable Drag-Based Editing on Multi-Modal Diffusion Transformers via Explicit Correspondence (ICLR 2026 · better on 4 of 4 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DiffEditor — DiffEditor: Boosting Accuracy and Flexibility on Diffusion-based Image Editing (CVPR 2024) · better on 6 of 6 shared results
- FastDrag — FastDrag: Manipulate Anything in One Step (NeurIPS 2024) · better on 6 of 6 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
- 3DGS-Drag: Dragging Gaussians for Intuitive Point-Based 3D Editing (ICLR 2025 · similarity 0.88)
- InstaInpaint: Instant 3D-Scene Inpainting with Masked Large Reconstruction Model (NeurIPS 2025 · similarity 0.87)
- Instant3dit: Multiview Inpainting for Fast Editing of 3D Objects (CVPR 2025 · similarity 0.87)
- FlowDrag: 3D-aware Drag-based Image Editing with Mesh-guided Deformation Vector Flow Fields (ICML 2025 · similarity 0.87)
- FastDrag: Manipulate Anything in One Step (NeurIPS 2024 · similarity 0.87)
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency (CVPR 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.
