Text2Video-Zero is a text-to-image diffusion model capable of zero-shot video generation.
Text2Video-Zero is an official implementation for zero-shot video generation using textual prompts or guidance from poses or edges, and Video Instruct-Pix2Pix. The repository includes code for all generation methods, and a low memory setup. Hyperparameters and optional ablation study can be defined. The code can be run using Gradio or Diffusers Library. Contributions are welcome. The license is CreativeML Open RAIL-M. If used in research, Text2Video-Zero should be cited.