Research map / Image generation / Editing and style transfer
Training-free Geometric Image Editing on Diffusion Models
Hanshen Zhu, Zhen Zhu, Kaile Zhang, Yiming Gong, Yuliang Liu, Xiang Bai
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 |
|---|---|---|---|
| decouple editing into transform, source inpainting, target refinement (FreeFine) rather than single-objective drag editing | FreeFine · GeoBench 2D edits | Warp error (WE): 0.098 -> 0.055 | Improved |
| apply FreeFine for 3D edits under the same depth-based transformation | FreeFine · GeoBench 3D edits | Mean distance (MD): 26.10 -> 20.32 | Improved |
| use training-free FreeFine refinement instead of trained BrushNet inpainting | FreeFine · GeoBench structure completion | FID: 186.93 -> 184.84 | Improved |
| FreeFine versus GeoDiffuser on 2D image quality | FreeFine · GeoBench 2D edits | FID: 33.89 vs 34.72 | No clear effect |
| use FreeFine on structure completion KD metric | FreeFine · GeoBench structure completion | KD: 0.971 -> 0.982 | Hurt |
| add Temporal Contextual Attention instead of MMSA, SSA, SDSA or plain self-attention (ablation) | FreeFine · GeoBench qualitative inpainting and refinement | qualitative quality | Improved |
| use TCA smooth transition instead of a single-threshold early stop of MMSA (ablation) | FreeFine · GeoBench qualitative | qualitative quality | Improved |
| add Local Perturbation (local DDPM) instead of deterministic DDIM (ablation) | FreeFine · GeoBench qualitative | qualitative quality | Improved |
| apply perturbation globally (Global Perturbation) instead of locally (ablation) | FreeFine · GeoBench qualitative | qualitative quality | Hurt |
| add Content-specified Generation with localized text guidance (ablation) | FreeFine · GeoBench qualitative | qualitative quality | Improved |
| apply guidance globally instead of within masks (ablation) | FreeFine · GeoBench qualitative | background and subject consistency | Hurt |
| use FreeFine versus DragonDiffusion in a user study | FreeFine · User study (perceptual quality) | preference rate: 20.8% -> 71.1% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DesignEdit — DesignEdit: Unify Spatial-Aware Image Editing via Training-free Inpainting with a Multi-Layered Latent Diffusion Framework (AAAI 2025) · better on 4 of 4 shared results
- BrushNet — BrushNet: A Plug-and-Play Image Inpainting Model with Decomposed Dual-Branch Diffusion (ECCV 2024) · better on 2 of 3 shared results
- RegionDrag — RegionDrag: Fast Region-Based Image Editing with Diffusion Models (ECCV 2024) · better on 1 of 1 shared result
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
- DesignEdit: Unify Spatial-Aware Image Editing via Training-free Inpainting with a Multi-Layered Latent Diffusion Framework (AAAI 2025 · similarity 0.88)
- Move Anything with Layered Scene Diffusion (CVPR 2024 · similarity 0.88)
- PAIR Diffusion: A Comprehensive Multimodal Object-Level Image Editor (CVPR 2024 · similarity 0.88)
- Diffusion Handles Enabling 3D Edits for Diffusion Models by Lifting Activations to 3D (CVPR 2024 · similarity 0.87)
- PixPerfect: Seamless Latent Diffusion Local Editing with Discriminative Pixel-Space Refinement (NeurIPS 2025 · similarity 0.87)
- DragDiffusion: Harnessing Diffusion Models for Interactive Point-based Image Editing (CVPR 2024 · 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.
