Research map / Video / Video generation
EditVerse: Unifying Image and Video Editing and Generation with In-Context Learning
Xuan Ju, Tianyu Wang, Yuqian Zhou, He Zhang, Qing Liu, Nanxuan Zhao, Zhifei Zhang, Yijun Li, Yuanhao Cai, Shaoteng Liu, Daniil Pakhomov, Zhe Lin and 2 more
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 |
|---|---|---|---|
| EditVerse unified model vs Lucy Edit/InsV2V instruction-guided editors | video editing · EditVerseBench | VLM editing quality: 5.89 -> 7.65 (vs Lucy Edit) | Improved |
| EditVerse vs commercial Runway Aleph on generation quality | video editing · EditVerseBench | Pick Score: 20.42 (Aleph) vs 20.07 | Hurt |
| add video editing training data (ablation) | EditVerse · EditVerseBench (20K steps) | VLM editing quality: 3.62 -> 6.95 | Improved |
| add image generation/editing data (ablation) | EditVerse · EditVerseBench (20K steps) | VLM editing quality: 6.40 -> 6.95 | Improved |
| add video generation data (ablation) | EditVerse · EditVerseBench (20K steps) | VLM editing quality: 6.52 -> 6.95 | Improved |
| add video generation data on temporal consistency (CLIP) (ablation) | EditVerse · EditVerseBench (20K steps) | CLIP consistency: 98.77 (w/o image) vs 98.68 (full) | No clear effect |
| add interleaved formulation (ablation) | EditVerse · EditVerseBench (20K steps) | VLM editing quality: 6.84 -> 6.95 | Improved |
| add sequential-dimension RoPE (ablation) | EditVerse · EditVerseBench (20K steps) | VLM editing quality: 6.42 -> 6.95 | Improved |
| design ablations on temporal consistency (ablation) | EditVerse · EditVerseBench | CLIP consistency: 98.69 -> 98.68 | No clear effect |
| apply EditVerse (trained without square video edits) to square TGVE+ videos | EditVerse · TGVE+ | ViCLIP_dir: 0.225 vs Movie Gen Edit 0.225 | No clear effect |
Later papers that compared with this method
5 later accepted papers report EditVerse as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- FFP-300K: Scaling First-Frame Propagation for Generalizable Video Editing (CVPR 2026 · better on 6 of 6 shared results)
- PropFly: Learning to Propagate via On-the-Fly Supervision from Pre-trained Video Diffusion Models (CVPR 2026 · better on 2 of 2 shared results)
- EasyV2V: A High-quality Instruction-based Video Editing Framework (CVPR 2026 · better on 2 of 2 shared results)
- LIVEditor-14B: Lightning Unified Video Editing via In-Context Sparse Attention (ICML 2026 · better on 1 of 1 shared result)
- EgoEdit: Dataset, Real-Time Streaming Model, and Benchmark for Egocentric Video Editing (CVPR 2026 · better on 0 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
- VINCIE-NExT: Unlocking Video Editing from Images via In-Context Modeling (NeurIPS 2026 · similarity 0.90)
- Señorita-2M: A High-Quality Instruction-based Dataset for General Video Editing by Video Specialists (NeurIPS 2025 · similarity 0.89)
- VicEdit: Learning to Edit Videos from Visual In-Context Examples (NeurIPS 2026 · similarity 0.89)
- ImgEdit: A Unified Image Editing Dataset and Benchmark (NeurIPS 2025 · similarity 0.89)
- Pathways on the Image Manifold: Image Editing via Video Generation (CVPR 2025 · similarity 0.88)
- EditBoard: Towards a Comprehensive Evaluation Benchmark for Text-Based Video Editing Models (AAAI 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.
