Research map / Video / Video generation
OpenVE-3M: A Large-Scale High-Quality Dataset for Instruction-Based Video Editing
Haoyang He, Jie Wang, Jiangning Zhang, Zhucun Xue, Xingyuan Bu, Qiangpeng Yang, Min Zheng, Lei Xie
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 |
|---|---|---|---|
| jointly feed video and instruction into an MLLM in addition to umT5 (ablation) | OpenVE-Edit · OpenVE-Bench (Seed1.6-VL judge) | overall score: 2.12 to 2.31 | Improved |
| add MoE-Connector over a shared-parameter connector (ablation) | OpenVE-Edit · OpenVE-Bench (Seed1.6-VL judge) | overall score: 2.31 to 2.41 | Improved |
| train on OpenVE-3M instead of Senorita-2M (ablation) | OpenVE-Edit · OpenVE-Bench (Seed1.6-VL judge) | overall score: 1.54 to 2.41 | Improved |
| use Canny edge control instead of depth control in data generation (ablation) | Wan2.1-Fun-Control data pipeline · data construction | qualitative detail | Improved |
| replace FlashAttention-2 with SageAttention-2 and cut steps from 50 to 10 (ablation) | Wan2.1-Fun-V1.1-14B-Control data generation · data construction | generation time: 50 minutes to 6 minutes | Improved |
| use MiniMax-Remover instead of DiffuEraser (ablation) | video removal data generation · data construction | inference time and consistency: twice the inference time | Mixed |
| score data with Seed1.6-VL rather than Qwen3-VL-30B-A3B as judge (ablation) | OpenVE-Bench · human alignment | PLCC: 0.583 to 0.833 | Improved |
| OpenVE-Edit (5B) versus 14B open-source baselines | OpenVE-Edit · OpenVE-Bench (Seed1.6-VL judge) | overall score: 2.41 vs 2.24 | 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.
- VideoCoF — VideoCoF: Unified Video Editing with Temporal Reasoner (CVPR 2026) · better on 2 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
- Señorita-2M: A High-Quality Instruction-based Dataset for General Video Editing by Video Specialists (NeurIPS 2025 · similarity 0.91)
- VEU-Bench: Towards Comprehensive Understanding of Video Editing (CVPR 2025 · similarity 0.89)
- HQ-Edit: A High-Quality Dataset for Instruction-based Image Editing (ICLR 2025 · similarity 0.89)
- ImgEdit: A Unified Image Editing Dataset and Benchmark (NeurIPS 2025 · similarity 0.89)
- VE-Bench: Subjective-Aligned Benchmark Suite for Text-Driven Video Editing Quality Assessment (AAAI 2025 · similarity 0.88)
- InsightEdit: Towards Better Instruction Following for Image Editing (CVPR 2025 · 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.
