Research map / Video / Video generation
InsViE-1M: Effective Instruction-based Video Editing with Elaborate Dataset Construction
Yuhui Wu, Liyi Chen, Ruibin Li, Shihao Wang, Chenxi Xie, Lei Zhang
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 GPT-4o filtering of training triplets (ablation) | InsViE training data · Test set (100 videos) | CLIP textual alignment: 18.47 -> 19.37 | Improved |
| add optical flow EPE filtering of training triplets (ablation) | InsViE training data · Test set (100 videos) | optical flow EPE: 6.58 -> 4.84 | Improved |
| add stage 2 training on higher-score, lower-EPE subset (Set-S2) (ablation) | InsViE after stage 1 · Test set (100 videos) | optical flow EPE: 5.08 -> 4.87 | Improved |
| add stage 3 (more static videos plus LPIPS loss) (ablation) | InsViE after stages 1&2 · Test set (100 videos) | DOVER: 0.519 -> 0.567 | Improved |
| add LPIPS loss in stage 3 (ablation) | InsViE stage 3 · Test set (100 videos) | DOVER: 0.543 -> 0.567 | Improved |
| add LPIPS loss in stage 3 (ablation) | InsViE stage 3 · Test set, temporal consistency GPT score | TC GPT score: 3.86 -> 3.88 | No clear effect |
| raise static:real video ratio in final stage from 0:1 to 5:1 (ablation) | InsViE stage 3 data · Test set (100 videos) | DOVER: 0.519 -> 0.567 | Improved |
| raise static:real ratio from 0:1 to 1:1 (ablation) | InsViE stage 3 data · Test set (100 videos) | CLIP textual alignment: 19.15 -> 19.18 | No clear effect |
| train InsV2V on InsViE-1M dataset (ablation) | InsV2V · TGVE | ViCLIP_dir: 0.208 -> 0.237 | Improved |
| use CogVideoX-based InsViE instead of retrained InsV2V on same data | InsViE-1M training · TGVE | ViCLIP_dir: 0.237 -> 0.242 | Improved |
| GPT-4o score alone as motion quality filter, without optical flow | data filtering · Test set (100 videos) | temporal consistency | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- RAVE — RAVE: Randomized Noise Shuffling for Fast and Consistent Video Editing with Diffusion Models (CVPR 2024) · better on 1 of 1 shared result
- Videoshop — Videoshop: Localized Semantic Video Editing with Noise-Extrapolated Diffusion Inversion (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
- Señorita-2M: A High-Quality Instruction-based Dataset for General Video Editing by Video Specialists (NeurIPS 2025 · similarity 0.91)
- Multi-Reward as Condition for Instruction-based Image Editing (ICLR 2025 · similarity 0.88)
- Instruct-CLIP: Improving Instruction-Guided Image Editing with Automated Data Refinement Using Contrastive Learning (CVPR 2025 · similarity 0.87)
- Learning Action and Reasoning-Centric Image Editing from Videos and Simulation (NeurIPS 2024 · similarity 0.86)
- VINCIE-NExT: Unlocking Video Editing from Images via In-Context Modeling (NeurIPS 2026 · similarity 0.86)
- InsightEdit: Towards Better Instruction Following for Image Editing (CVPR 2025 · 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.
