The idea of prompt “engineering” strikes me as nonsense. I’ve spent many hours on Midjourney, and it makes some astonishing artworks. But it essentially never makes what you expect. I see that as a feature rather than a bug. Instead of professional prompt engineers who somehow know the precise incantation to generate exactly the picture that’s in your head (they don’t), I think the actual future direction of this business is large numbers of people skimming through thousands of fairly similar images and varying the preferred ones over and over to inch towards the desired output. Of course, all that will be obviated when the AI itself advances by another big step, anyhow.