My two cents: the techniques OP uses are absolutely valid, but I've found much more success "sampling" styles and poses from existing works.
Rather than trying to perfectly describe my image, I like to use references where the source material has what you want. With minimal direction these prompts get impressively close:
I sunk ~20 hours and $100 playing with DALL-E since last week and I've had a very different experience. Sure--my first dozen attempts with the engine gave bad results, but once I learned to "speak its language" it got easier to generate highly-polished images. The most realistic results come by appending things like "realistic photograph, 4k, in the style of a fashion magazine" to prompts. I suppose any style would work, as long as the body of source material in that style is (mostly) high-quality.
Here's a couple examples I produced with just a little trial and error. FWIW I have an engineering background and zero design experience.
Maybe they're not perfect, but I'm impressed as hell. Exploring what's possible by wording prompts differently feels very much like using a search engine for the first time. Give it a year. This technology is going places.