SkyReels-V2: Infinite-length Film Generative Model
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908328124219392 |
|---|---|
| author | Chen, Guibin Lin, Dixuan Yang, Jiangping Lin, Chunze Zhu, Junchen Fan, Mingyuan Zhang, Hao Chen, Sheng Chen, Zheng Ma, Chengcheng Xiong, Weiming Wang, Wei Pang, Nuo Kang, Kang Xu, Zhiheng Jin, Yuzhe Liang, Yupeng Song, Yubing Zhao, Peng Xu, Boyuan Qiu, Di Li, Debang Fei, Zhengcong Li, Yang Zhou, Yahui |
| author_facet | Chen, Guibin Lin, Dixuan Yang, Jiangping Lin, Chunze Zhu, Junchen Fan, Mingyuan Zhang, Hao Chen, Sheng Chen, Zheng Ma, Chengcheng Xiong, Weiming Wang, Wei Pang, Nuo Kang, Kang Xu, Zhiheng Jin, Yuzhe Liang, Yupeng Song, Yubing Zhao, Peng Xu, Boyuan Qiu, Di Li, Debang Fei, Zhengcong Li, Yang Zhou, Yahui |
| contents | Recent advances in video generation have been driven by diffusion models and autoregressive frameworks, yet critical challenges persist in harmonizing prompt adherence, visual quality, motion dynamics, and duration: compromises in motion dynamics to enhance temporal visual quality, constrained video duration (5-10 seconds) to prioritize resolution, and inadequate shot-aware generation stemming from general-purpose MLLMs' inability to interpret cinematic grammar, such as shot composition, actor expressions, and camera motions. These intertwined limitations hinder realistic long-form synthesis and professional film-style generation. To address these limitations, we propose SkyReels-V2, an Infinite-length Film Generative Model, that synergizes Multi-modal Large Language Model (MLLM), Multi-stage Pretraining, Reinforcement Learning, and Diffusion Forcing Framework. Firstly, we design a comprehensive structural representation of video that combines the general descriptions by the Multi-modal LLM and the detailed shot language by sub-expert models. Aided with human annotation, we then train a unified Video Captioner, named SkyCaptioner-V1, to efficiently label the video data. Secondly, we establish progressive-resolution pretraining for the fundamental video generation, followed by a four-stage post-training enhancement: Initial concept-balanced Supervised Fine-Tuning (SFT) improves baseline quality; Motion-specific Reinforcement Learning (RL) training with human-annotated and synthetic distortion data addresses dynamic artifacts; Our diffusion forcing framework with non-decreasing noise schedules enables long-video synthesis in an efficient search space; Final high-quality SFT refines visual fidelity. All the code and models are available at https://github.com/SkyworkAI/SkyReels-V2. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_13074 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SkyReels-V2: Infinite-length Film Generative Model Chen, Guibin Lin, Dixuan Yang, Jiangping Lin, Chunze Zhu, Junchen Fan, Mingyuan Zhang, Hao Chen, Sheng Chen, Zheng Ma, Chengcheng Xiong, Weiming Wang, Wei Pang, Nuo Kang, Kang Xu, Zhiheng Jin, Yuzhe Liang, Yupeng Song, Yubing Zhao, Peng Xu, Boyuan Qiu, Di Li, Debang Fei, Zhengcong Li, Yang Zhou, Yahui Computer Vision and Pattern Recognition Recent advances in video generation have been driven by diffusion models and autoregressive frameworks, yet critical challenges persist in harmonizing prompt adherence, visual quality, motion dynamics, and duration: compromises in motion dynamics to enhance temporal visual quality, constrained video duration (5-10 seconds) to prioritize resolution, and inadequate shot-aware generation stemming from general-purpose MLLMs' inability to interpret cinematic grammar, such as shot composition, actor expressions, and camera motions. These intertwined limitations hinder realistic long-form synthesis and professional film-style generation. To address these limitations, we propose SkyReels-V2, an Infinite-length Film Generative Model, that synergizes Multi-modal Large Language Model (MLLM), Multi-stage Pretraining, Reinforcement Learning, and Diffusion Forcing Framework. Firstly, we design a comprehensive structural representation of video that combines the general descriptions by the Multi-modal LLM and the detailed shot language by sub-expert models. Aided with human annotation, we then train a unified Video Captioner, named SkyCaptioner-V1, to efficiently label the video data. Secondly, we establish progressive-resolution pretraining for the fundamental video generation, followed by a four-stage post-training enhancement: Initial concept-balanced Supervised Fine-Tuning (SFT) improves baseline quality; Motion-specific Reinforcement Learning (RL) training with human-annotated and synthetic distortion data addresses dynamic artifacts; Our diffusion forcing framework with non-decreasing noise schedules enables long-video synthesis in an efficient search space; Final high-quality SFT refines visual fidelity. All the code and models are available at https://github.com/SkyworkAI/SkyReels-V2. |
| title | SkyReels-V2: Infinite-length Film Generative Model |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2504.13074 |