xGen-VideoSyn-1: High-fidelity Text-to-Video Synthesis with Compressed Representations
Research paper by Can Qin and collaborators. Published 2024-08-22. Topics: latent diffusion model, VVVAE, spatial compression, temporal compression, visual tokens.
Resource information
- Authors
- Can Qin, Congying Xia, Krithika Ramakrishnan, Michael Ryoo, Lifu Tu, Yihao Feng, Manli Shu, Honglu Zhou, Anas Awadalla, Jun Wang, Senthil Purushwalkam, Le Xue, Yingbo Zhou, Huan Wang, Silvio Savarese, Juan Carlos Niebles, Zeyuan Chen, Ran Xu, Caiming Xiong
- Research topics
- latent diffusion model, VVVAE, spatial compression, temporal compression, visual tokens, computational demands, divide-and-merge strategy, temporal consistency, Diffusion Transformer, spatial self-attention, temporal self-attention, robust generalization
- Publication date
- 2024-08-22
- Publication
- arXiv
- Last record update
- Nov 3, 2026
