Img2Img capabilities? #23
Description
I tried to add img2img in the pipeline as well as in the comfyUI nodes and noticed that it doesn't work as expected. Was the model not trained on img2img? I looked into inverse DDIM as well, but it'd take way too much vram.