/MovieDreamer

MovieDreamer: Hierarchical Generation for Coherent Long Visual Sequences

📖 Project Page | 📄 Paper Link | 🤗 Huggingface Paper Link | 📹 Demo

MovieDreamer aims to provide a solution for long content generation. Our method is capable of generating both long story and long video with multiple characters well-preserved!

📣 News

  • [7/20/2024] Paper released!

😲 Results

Please refer to our homepage for more thrilling results!

🛠️ Setup

  • Comming Soon...

🚩 Plan

  • Story Results and Video Results.
  • Paper Released.
  • Source Code and Pretrained Weights of Diffusion AutoEncoder.
  • Source Code and Pretrained Weights of Autoregressive Model.
  • Source code of Video Generation.
  • Source code of Data Pre-processing.
  • Source code of Training.

🎫 License

For academic use, this project is licensed under the 2-clause BSD License. For commercial use, please contact Chunhua Shen.

🖊️ Citation

@misc{zhao2024moviedreamerhierarchicalgenerationcoherent,
      title={MovieDreamer: Hierarchical Generation for Coherent Long Visual Sequence}, 
      author={Canyu Zhao and Mingyu Liu and Wen Wang and Jianlong Yuan and Hao Chen and Bo Zhang and Chunhua Shen},
      year={2024},
      eprint={2407.16655},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2407.16655}, 
}