LaVie: high-quality video generation with cascaded latent diffusion models

• Main Authors: Wang, Yaohui, Chen, Xinyuan, Ma, Xin, Zhou, Shangchen, Huang, Ziqi, Wang, Yi, Yang, Ceyuan, He, Yinan, Yu, Jiashuo, Yang, Peiqing, Guo, Yuwei, Wu, Tianxing, Si, Chenyang, Jiang, Yuming, Chen, Cunjian, Loy, Chen Change, Dai, Bo, Lin, Dahua, Qiao, Yu, Liu, Ziwei
• 其他作者: College of Computing and Data Science
• 格式: Article
• 語言: English
• 出版: 2025
• 主題: Computer and Information Science; Diffusion models; Generative modeling
• 在線閱讀: https://hdl.handle.net/10356/183061

This work aims to learn a high-quality text-to-video (T2V) generative model by leveraging a pre-trained text-to-image (T2I) model as a basis. It is a highly desirable yet challenging task to simultaneously (a) accomplish the synthesis of visually realistic and temporally coherent videos while (b) pr...