I was pretty impressed with Flux, plan to use it more. For a Tarot deck, which has a bunch of nude figures, it was pretty determined to clothe them; I eventually just left the woman in The World wearing something. Describing the image as "NSFW" helped; I'm sure that people have their own techniques that I just don't know about.
I'm used to being able to use regional prompting in Stable Diffusion to stick specific things at specific places in the image. I don't know yet if there's a regional prompting analog compatible with Flux; the Stable Diffusion and Flux workflows are (unexpectedly to me) quite different in ComfyUI. Flux does understand some level of English-like description of the layout of the image, which is cool, but I wasn't always able to get the output I wanted with that, so I expect that there's still more digging.
Yeah, he's got a few quirks. Flux didn't like doing things upside-down much. May be a good way to go about doing that (other than just generating and flipping the image, which has drawbacks of its own), but I haven't gone investigating.