Why doesn't Microsoft Designer (AI images) allow "revealing" requests, but creates totally revealing images itself?
It's kind of paradoxical. If I write that someone is standing by a lake in swimwear, they say it's against the rules.
But then I write, for example, a woman is hiking in the mountains. She's wearing a wool sweater. — And then the woman in the pictures is wearing ONLY a wool sweater and showing me her butt. And I think to myself: That's much more offensive than someone in swimwear at a lake.
Because the AI was probably made by the developers regulations, which they “required”, but it is still so unripe/errorious/… that they do not recognize the naked Po as such, but only as a collection of colorful pixels. 😅 My guess.
I personally have made the experience that the pictures make what they want. Mostly, the image corresponds to what one has imagined, but also only approximately. Writing errors often occur, e.g. in the case of desired letterings or signs, in addition, fables are often made from animals (e.g. a fox head on a bird body in orange color).
Probably it is possible to fix a lot by a special formulation, but I don’t know how well. I try to describe the desired image as simple as possible, but in detail, if necessary. Maybe that’s wrong.