When I was at a convenience store, I saw someone dressed as a bear clown. I photographed him from a distance. On a whim, I uploaded the photo to the DALLE-2 site. I generate variations without prompt. And unexpectedly, the result is like this.
Looks, right? The result expression of the DALLE-2 generation,shows feelings of sadness, either tiredness, or a sense of pity.
What do you think? Apart from DALLE-2's rule which is not allowed to upload real photo faces, I think this is just an assumption of the AI algorithm.