Research map / Video / Video generation
OneStory: Coherent Multi-Shot Video Generation with Adaptive Memory
Zhaochong An, Menglin Jia, Haonan Qiu, Zijian Zhou, Xiaoke Huang, Zhiheng Liu, Weiming Ren, Kumara Kahatapitiya, Ding Liu, Sen He, Chenyang Zhang, Tao Xiang and 3 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 |
|---|---|---|---|
| add both Adaptive Conditioner and Frame Selection to last-frame-only conditioning (ablation) | OneStory (Wan2.1 I2V) · Text-to-multi-shot ablation | Character consistency: 0.5153 -> 0.5874 | Improved |
| add both Adaptive Conditioner and Frame Selection to last-frame-only conditioning (ablation) | OneStory (Wan2.1 I2V) · Text-to-multi-shot ablation | Semantic alignment: 0.1814 -> 0.2389 | Improved |
| add shot inflation (inflate two-shot clips to three shots) (ablation) | OneStory training · Text-to-multi-shot ablation | Character consistency: 0.5514 -> 0.5649 | Improved |
| add decoupled conditioning curriculum on top of shot inflation (ablation) | OneStory training · Text-to-multi-shot ablation | Character consistency: 0.5649 -> 0.5874 | Improved |
| add decoupled conditioning curriculum on top of shot inflation (ablation) | OneStory training · Text-to-multi-shot ablation | Environment consistency: 0.5790 -> 0.5752 | Hurt |
| increase context tokens from one to three latent-frame equivalents (ablation) | OneStory Adaptive Conditioner · Text-to-multi-shot ablation | Environment consistency: 0.5752 -> 0.5863 | Improved |
| increase context tokens from one to three latent-frame equivalents (ablation) | OneStory Adaptive Conditioner · Text-to-multi-shot ablation | Character consistency: 0.5874 -> 0.5926 | Improved |
| use automatic frame selection instead of uniform or most-recent sampling (ablation) | OneStory · Qualitative frame selection comparison | contextual consistency | Improved |
| use OneStory instead of fixed-window Mask2DiT | multi-shot generation · Text-to-multi-shot | Character consistency: 0.5472 -> 0.5874 | Improved |
| use OneStory instead of keyframe conditioning (StoryDiffusion + Wan2.1) | multi-shot generation · Text-to-multi-shot | Semantic alignment: 0.2217 -> 0.2389 | 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.
- Mask2DiT — Mask^2DiT: Dual Mask-based Diffusion Transformer for Multi-Scene Long Video Generation (CVPR 2025) · better on 4 of 4 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
- ShotStream: Streaming Multi-Shot Video Generation for Interactive Storytelling (ECCV 2026 · similarity 0.91)
- EntityBench: Towards Entity-Consistent Long-Range Multi-Shot Video Generation (NeurIPS 2026 · similarity 0.88)
- MovieDreamer: Hierarchical Generation for Coherent Long Visual Sequences (ICLR 2025 · similarity 0.88)
- ShotAdapter: Text-to-Multi-Shot Video Generation with Diffusion Models (CVPR 2025 · similarity 0.88)
- CineOrchestra: Unified Entity-Centric Conditioning for Cinematic Video Generation (NeurIPS 2026 · similarity 0.87)
- SEINE: Short-to-Long Video Diffusion Model for Generative Transition and Prediction (ICLR 2024 · similarity 0.87)
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.
