Feature request: Image to image

[mortenjust]

Would be great to be able to prompt the model with a prompt along with an image

[Kitenite]

Yes, I would second this. I got this to run on an iPad Pro 4th generation (3.1s/step). I'm so thrilled that this is working!