Research map / Video / Video generation
OSVE: One Step Video Editing with One Step Diffusion Models
Habin Lim, Gyeong-Moon Park
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 SAE loss to reconstruction loss for encoder training (ablation) | OSVE encoder · PIE-Bench | structure distance: 0.087 -> 0.064 | Improved |
| add SAE loss to reconstruction loss (ablation) | OSVE encoder · PIE-Bench | CLIP Edit: 19.884 -> 20.416 | Improved |
| add reconstruction loss to SAE loss (ablation) | OSVE encoder · PIE-Bench | CLIP Whole: 22.349 -> 22.329 | No clear effect |
| add sliding window to UFE (ablation) | OSVE · 90-frame videos | Subject Consistency: 0.931 -> 0.954 | Improved |
| add anchor frame to UFE (ablation) | OSVE · 90-frame videos | Subject Consistency: 0.931 -> 0.943 | Improved |
| combine sliding window and anchor (ablation) | OSVE · 90-frame videos | BQS: 0.671 -> 0.677 | Improved |
| apply sliding window and anchor (UFE long video) (ablation) | OSVE · 90-frame videos | Aesthetic Quality: 0.680 -> 0.670 | Hurt |
| use one-step OSVE instead of multi-step methods | OSVE · short videos | BQS: 0.674 -> 0.679 (vs FRESCO) | Improved |
| use one-step OSVE instead of RAVE | OSVE · short videos | FPS: 0.091 -> 15.625 | Improved |
| use one-step OSVE instead of RAVE | OSVE · long videos | FPS: 0.102 -> 15.793 | Improved |
| use OSVE vs best baseline in user study | OSVE · user study | temporal consistency: 3.68 -> 3.85 | Improved |
| OSVE vs Plug-and-Play and FRESCO on single metrics | OSVE · short videos | Temporal Flickering / Imaging Quality: TF 0.995 (PnP) vs 0.978; IQ 0.729 (FRESCO) vs 0.703 | Mixed |
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.
- FRESCO — FRESCO: Spatial-Temporal Correspondence for Zero-Shot Video Translation (CVPR 2024) · better on 4 of 4 shared results
- RAVE — RAVE: Randomized Noise Shuffling for Fast and Consistent Video Editing with Diffusion Models (CVPR 2024) · better on 4 of 4 shared results
- COVE — COVE: Unleashing the Diffusion Feature Correspondence for Consistent Video Editing (NeurIPS 2024) · 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
- COVE: Unleashing the Diffusion Feature Correspondence for Consistent Video Editing (NeurIPS 2024 · similarity 0.89)
- Visual Prompting for One-shot Controllable Video Editing without Inversion (CVPR 2025 · similarity 0.89)
- Transforming Image Editors into Video Editors (NeurIPS 2026 · similarity 0.89)
- VideoDirector: Precise Video Editing via Text-to-Video Models (CVPR 2025 · similarity 0.89)
- FADE: Frequency-Aware Diffusion Model Factorization for Video Editing (CVPR 2025 · similarity 0.89)
- VINCIE-NExT: Unlocking Video Editing from Images via In-Context Modeling (NeurIPS 2026 · similarity 0.88)
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.
