I see the rules currently state “Nothing from realistic models trying to generate real people.”
To verify I understand this correctly, does this mean realistic imagery is OK as long as it does not mimic the likeness of a real person (aka deepfakes)?
For example, a photo-realistic image of a body builder is OK, but a photo-realistic image of Arnold Schwarzenegger is not?
Yeah that one’s pretty convincing. The scale of the fruit is way off (not even grapefruit is that big), but that’s about all I can see. It was a good call to have her hands behind her head. I have a suspicion that if the image were higher resolution that it’d reveal her teeth were a mess. EDIT: the shape of her eyes doesn’t exactly match, but that could be subjective.
For the second one (on the same page), there are several tells: There’s a giant lemon that somehow has an apple stem (or that’s how it looks to me). The fingers on her hands are messed up. One hand has too many fingers, which creeps me out. Of course my choice in art probably creeps people out already, so… moving on. Her hair gets mixed in with the top edge of the towel she’s lying on. You can see how it somehow extends the towel upward in places, almost like you’d see with a selective blur tool. This kind of thing seems to happen a lot with complex boundaries in AI images. Messy hair in particular often has problems.
EDIT: Oh, and I got carried away and forgot to write this, but: thanks for the review!