By switching to diffusers, this has no future. Animatediff uses diffusers but lacks the compatibility with Stable Diffusion users who use the Automatic1111 or ComfyUI webUIs as these implementations use the standard codebase and the result is inconsistent output between text to image stills vs animations (diffusers). E.g. seeds essentially don't match up.
If the pre-diffusers code of Text2Video-Zero can result in consistent seeds between text2image, T2VZ will easily become the standard over Animatediff.
By switching to diffusers, this has no future. Animatediff uses diffusers but lacks the compatibility with Stable Diffusion users who use the Automatic1111 or ComfyUI webUIs as these implementations use the standard codebase and the result is inconsistent output between text to image stills vs animations (diffusers). E.g. seeds essentially don't match up.
If the pre-diffusers code of Text2Video-Zero can result in consistent seeds between text2image, T2VZ will easily become the standard over Animatediff.