A finetuned version of T2V, with better understanding of prompts and dynamic video generation.
This model only supports ComfyUI generation
Recommended sampler settings:
Fast Generation
- Steps: 10
- Sampler: Heun
- CGF: 1
- Scheduler: beta57
Best Result
- Steps: 30
- Sampler: Eular
- CGF: 1
- Scheduler: normal