Research map / Video / Video generation
Video Editing via Factorized Diffusion Distillation
Uriel Singer, Amit Zohar, Yuval Kirstain, Shelly Sheynin, Adam Polyak, Devi Parikh, Yaniv Taigman
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 Factorized Diffusion Distillation to align edit and video adapters (EVE vs InsV2V) | EVE · TGVE | ViCLIP_dir: 0.208 -> 0.221 | Improved |
| EVE compared with STDF on CLIPFrame | EVE · TGVE | CLIPFrame: 0.933 -> 0.922 | Hurt |
| EVE compared with baselines in human evaluation | EVE · TGVE | human preference for EVE, average (%): 59.6 (vs InsV2V) to 77.2 (vs TAV) | Improved |
| add alignment of pre-trained adapters via FDD (ablation) | EVE · TGVE (human preference) | human preference for EVE, average (%): 86.3 preferred EVE over w/o alignment | Improved |
| initialize with pre-trained adapters instead of random init and joint learning (ablation) | EVE · TGVE (human preference) | human preference for EVE, average (%): 87.2 preferred EVE over Random Init | Improved |
| add SDS loss to adversarial loss (ablation) | EVE · TGVE (human preference) | human preference for EVE, average (%): 85.7 preferred EVE over w/o SDS | Improved |
| add adversarial discriminators to SDS loss (ablation) | EVE · TGVE (human preference) | human preference for EVE, average (%): 80.8 preferred EVE over w/o discriminators | Improved |
| sample diffusion steps from k bins instead of uniformly (ablation) | EVE · TGVE (human preference) | human preference for EVE, average (%): 53.0 preferred EVE over w/o K-Bin Sampling | Improved |
Closest papers by meaning
- FADE: Frequency-Aware Diffusion Model Factorization for Video Editing (CVPR 2025 · similarity 0.88)
- Transforming Image Editors into Video Editors (NeurIPS 2026 · similarity 0.87)
- CCEdit: Creative and Controllable Video Editing via Diffusion Models (CVPR 2024 · similarity 0.87)
- VINCIE-NExT: Unlocking Video Editing from Images via In-Context Modeling (NeurIPS 2026 · similarity 0.87)
- Ground-A-Video: Zero-shot Grounded Video Editing using Text-to-image Diffusion Models (ICLR 2024 · similarity 0.86)
- VidToMe: Video Token Merging for Zero-Shot Video Editing (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.
