Support HuggingFace Diffusers

#4
by oxysoft - opened

It seems that the repository is missing a model_index or doesn't have the proper folder structure for use with HuggingFace Diffusers.

Diffusers' ControlNetModel currently doesn't support loading a ControlNet that has an input with different number of channels than 3.

I've made a tiny change (see PR here) that fixes this though.

I've pushed the converted Diffusers checkpoint to this repo here: https://huggingface.co./wav/TemporalNet2

There's a script in there that'll run using HuggingFace Diffusers if you install my branch:

pip install git+https://github.com/JCBrouwer/diffusers@expose-controlnet-conditioning-channels
python temporalvideo_hf.py --help

Sign up or log in to comment