VideoPoet

大型语言模型

VideoPoet是由 Google Research 于 2023 年开发的一款大型语言模型,主要用于影片制作。[1][2][3][4] 该模型能将静态影像转换为动画。[5] VideoPoet 支持文本、影像和影片作为输入,并能将这些输入转换成多种格式。[4] 该模型于 2023 年 12 月 19 日正式公开。[1]VideoPoet 使用自我回归模型

VideoPoet
"一只狗在电影院里吃爆米花"
"一只戴着帽子、太阳眼镜和皮夹克的泰迪熊正在打鼓"
由该模型生成的示例影片来自于文本
开发者Google
首次发布2024年2月8日,​9个月前​(2024-02-08
类型大型语言模型

参考资料

  1. ^ 1.0 1.1 Krithika, K. L. Google Unveils VideoPoet, a New LLM for Video Generation. Analytics India Magazine. 2023-12-20 [2024-04-29] (美国英语). 
  2. ^ Kondratyuk, Dan; Yu, Lijun; Gu, Xiuye; Lezama, José; Huang, Jonathan; Hornung, Rachel; Adam, Hartwig; Akbari, Hassan; Alon, Yair; Birodkar, Vighnesh; Cheng, Yong; Chiu, Ming-Chang; Dillon, Josh; Essa, Irfan; Gupta, Agrim; Hahn, Meera; Hauth, Anja; Hendon, David; Martinez, Alonso; Minnen, David; Ross, David; Schindler, Grant; Sirotenko, Mikhail; Sohn, Kihyuk; Somandepalli, Krishna; Wang, Huisheng; Yan, Jimmy; Yang, Ming-Hsuan; Yang, Xuan; Seybold, Bryan; Jiang, Lu. VideoPoet: A Large Language Model for Zero-Shot Video Generation. December 21, 2023. arXiv:2312.14125  [cs.CV]. 
  3. ^ Google has introduced VideoPOET breaking new ground in coherent video generation. Gizmochina. December 21, 2023. 
  4. ^ 4.0 4.1 VideoPoet. Google Research. [2024-04-29] (英语). 
  5. ^ Franzen, Carl. Google’s new multimodal AI video generator VideoPoet looks incredible. VentureBeat. December 20, 2023. 

外部链接