For Viking man I used gpt-image-2 through API, hence why I didn’t ask the model. Although when I prompted the second time “A Viking man”, I received an output. A bit weird.
And this was even weirder, I used ChatGPT and prompted “Viking woman” without any probs.
And yes, I see what you mean about the Vikings TV-show look. My guess is that this kind of visual stereotype is strongly represented in the model’s training patterns, so “Viking man/woman” tends to drift toward that familiar dark, rugged, cinematic look unless the prompt pushes it somewhere else.
Might be stupid question here @_j, is that a real show? Remember I’m not from the States
Tiny challenge for those that are up for it: save/download this image, then edit it so that the right flag goes on the right European country and then post it here. And please try not to cheat