Player_Diff said:
Fully agreed. Also in post https://danbooru.donmai.us/posts/6702410 its not even the hairclip alone, her ear, her hair (especially by her hand, I have honestly no clue what is even happening there behind), and her left arm by her the torn clothes. I'm farily certain and confident that those are AI-generated.
Something I noted is that in post #6702410, you can see line-art artifacts. So this was originally line-art presumably drawn by a human. Notice the hair on the right side of the face that the AI has failed to colour in correctly. Some of the thin strands of hair were obviously added in after the AI had coloured the image, like the ones that terminate near the cleavage; they are too thin for the AI to have created, and they lack the characteristic black shading pattern present in the rest of the image. You can also see some of the original line-art around the eyebrows. This is consistent with image-to-image generation, since a de-noising value needs to be defined for the whole image at once, but the AI will not alter the image uniformly, sometimes leaving small regions of the original image essentially intact.
You can notice sort of the same thing going on in post #6702411, where the AI has duplicated the nose and eyebrows, but left the original lines somewhat intact. This one seems to have been more "processed" by the AI than the other one. That lines up with the higher resolution on the second image, since it takes a long time to run these image-to-image transformations, especially at high resolutions. So in this image the original lineart is mostly gone at this point.
I think that this might as well be AI art, though. The original sketch was obviously a messy draft, nothing more, as you can see from the rather ugly leftover lines in post #6702410. The image that came out the other side is basically just an AI generated image that was placed into a particular pose using the sketch instead of word prompts. The eyes are muddy, there are way too many drips, the clothes and anatomy are nonsense. It can't decide whether the white stuff should be on the subject or the camera lens. So I believe that it definitely does deserve the Ai-generated tag. I just wanted to note that it does look like the artist was at least being truthful, and this is, in the loosest possible sense, a "automatic colouring" of something made by an actual human.