Almost everything was done solely within the bounds of the stable diffusion program with the exceptions of:
some resizing.
minor vibrancy/color correcting on the eye.
denoising script.
creation of an upload mask.
simple shapes were painted to re-add the front leg and to extend the neck/body area.
I think most people would agree that resizing and a denoise script doesn’t change it to be ‘assisted’, so its how you see the leg/neck thing. Personally since it was like 1 minutes of work it doesn’t seem right to pretend the AI was second fiddle.
Hell, you could argue that all inpainting is ‘AI assisted’ because its ultimately a human stitching a scene together. I don’t know the answer here, so I went for the simplest tag.
Yeah, I guess it’s too little human input to qualify for that tag.
Frankly, I don’t know.
So is this still machine learning generated, or is this merely machine learning assisted?
I see. Yeah, I can tell a lot of curation went into this :)
It is almost entirely inpainting, so the prompt isn’t very useful.
https://streamable.com/zuf69v