Research map / Image generation / Editing and style transfer
V-HOLD: Stabilizing Flow Trajectories to Rethink the Edit–Preservation Trade-off
Gahyeon Kim, Dong-oh Kang
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 V-HOLD direction holding to FlowEdit | FlowEdit (FLUX.1-dev) · PIE-Bench | DINO Dist_src: 0.8004 -> 0.7862 | Improved |
| add V-HOLD to FlowEdit | FlowEdit (SD3-medium) · PIE-Bench | DINO Dist_src: 0.8873 -> 0.8461 | Improved |
| add V-HOLD to FlowEdit | FlowEdit (SD3-medium) · FlowEdit-Data | DINO Dist_src: 0.8831 -> 0.7547 | Improved |
| add V-HOLD to FlowEdit | FlowEdit (FLUX.1-dev) · PIE-Bench | LPIPS_src: 0.2712 -> 0.2874 | Hurt |
| add V-HOLD to FlowEdit | FlowEdit (SD3-medium) · FlowEdit-Data | LPIPS_src: 0.2243 -> 0.2378 | Hurt |
| add V-HOLD to FlowEdit | FlowEdit (FLUX.1-dev) · PIE-Bench | CLIP_tgt: 0.2583 -> 0.2623 | Improved |
| add V-HOLD to FlowEdit | FlowEdit (SD3-medium) · PIE-Bench | CLIP_tgt: 0.2781 -> 0.2723 | Hurt |
| add V-HOLD to FlowEdit | FlowEdit (SD3-medium) · trajectory stability | Instability: 0.636 -> 0.090 (-85.8%) | Improved |
| add V-HOLD to FlowEdit | FlowEdit (FLUX.1-dev) · trajectory stability | Update volatility: 0.315 -> 0.223 (-29.3%) | Improved |
| add V-HOLD to FlowEdit | FlowEdit (FLUX.1-dev) · efficiency | NFE: 24 -> 21 | Improved |
| add V-HOLD to FlowEdit | FlowEdit (SD3-medium) · efficiency | edit time (s): 3.420 -> 2.177 | Improved |
| add candidate sampling (K=3) to holding (ablation) | V-HOLD (FLUX.1-dev) · PIE-Bench | DINO Dist_src: 0.992 -> 0.786 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- FlowEdit — FlowEdit: Inversion-Free Text-Based Editing Using Pre-Trained Flow Models (ICCV 2025) · better on 7 of 9 shared results
- FlowAlign — FlowAlign: Trajectory-Regularized, Inversion-Free Flow-based Image Editing (ICLR 2026) · better on 2 of 2 shared results
- FireFlow — FireFlow: Fast Inversion of Rectified Flow for Image Semantic Editing (ICML 2025) · better on 2 of 2 shared results
- SDEdit — SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations (ICLR 2022) · better on 2 of 2 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
- Constrained Look-ahead Guidance for Interference-Aware Flow Editing (NeurIPS 2026 · similarity 0.91)
- Why Do DiT Editors Drift? Plug-and-Play Low Frequency Alignment in VAE Latent Space (NeurIPS 2026 · similarity 0.88)
- SplitFlow: Flow Decomposition for Inversion-Free Text-to-Image Editing (NeurIPS 2025 · similarity 0.88)
- Preserve or Modify? Context-Aware Evaluation for Balancing Preservation and Modification in Text-Guided Image Editing (CVPR 2025 · similarity 0.87)
- Improving Editability in Image Generation with Layer-wise Memory (CVPR 2025 · similarity 0.87)
- Inversion-Free Image Editing with Language-Guided Diffusion Models (CVPR 2024 · 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.
