Research map / Image generation / Editing and style transfer
h-Flow: Flexible Flow-based Image Editing via Doob's h-Transform
ZeHui Guo, Zhen Wang, Junwei Shu, Changbo Wang, Long Chen, Yang Li
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 |
|---|---|---|---|
| add orthogonal projection of the editing direction (ablation) | h-Flow · PIE-Bench | CLIP-Entire: 25.37 -> 25.80 | Improved |
| add orthogonal projection of the editing direction (ablation) | h-Flow · PIE-Bench | SSIM: 0.834 -> 0.840 | Improved |
| add orthogonal projection of the editing direction (ablation) | h-Flow · PIE-Bench | structure distance: 0.026 -> 0.026 | No clear effect |
| increase editing strength lambda_edit from 0 to 1.5 (ablation) | h-Flow · PIE-Bench | CLIP-Edited: 20.49 -> 22.80 | Improved |
| increase editing strength lambda_edit from 1.5 to 3.0 (ablation) | h-Flow · PIE-Bench | SSIM: 0.840 -> 0.804 | Mixed |
| increase editing strength lambda_edit from 1.5 to 3.0 (ablation) | h-Flow · PIE-Bench | CLIP-Edited: 22.80 -> 23.49 | Mixed |
| increase reconstruction strength lambda_rec from 0 to 1.5 (ablation) | h-Flow · PIE-Bench | SSIM: 0.830 -> 0.856 | Improved |
| increase reconstruction strength lambda_rec from 0 to 1.5 (ablation) | h-Flow · PIE-Bench | CLIP-Edited: 22.92 -> 22.73 | No clear effect |
| add h-Flow to RF-Inversion forward process (ablation) | RF-Inversion · PIE-Bench | CLIP-Edited: +1.26 | Improved |
| add h-Flow to SDEdit forward process (ablation) | SDEdit · PIE-Bench | CLIP-Edited: +0.56 | Improved |
| add h-Flow to ODE-Inv forward process (ablation) | ODE-Inv · PIE-Bench | SSIM: 0.62 -> 0.70 | Improved |
| add h-Flow to ODE-Inv forward process (ablation) | ODE-Inv · PIE-Bench | structure distance: 0.130 -> 0.077 | Improved |
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.
- FlowEdit — FlowEdit: Inversion-Free Text-Based Editing Using Pre-Trained Flow Models (ICCV 2025) · better on 3 of 3 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
- SplitFlow: Flow Decomposition for Inversion-Free Text-to-Image Editing (NeurIPS 2025 · similarity 0.91)
- h-Edit: Effective and Flexible Diffusion-Based Editing via Doob's h-Transform (CVPR 2025 · similarity 0.89)
- Unveil Inversion and Invariance in Flow Transformer for Versatile Image Editing (CVPR 2025 · similarity 0.89)
- Latent Space Editing in Transformer-Based Flow Matching (AAAI 2024 · similarity 0.87)
- Stable Flow: Vital Layers for Training-Free Image Editing (CVPR 2025 · similarity 0.87)
- HiFlow: Training-free High-Resolution Image Generation with Flow-Aligned Guidance (NeurIPS 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.
