"Tiny living woman in red leggings and a brown hooded sweatshirt, standing on her desk, looking up at the monitor."
Will get me this image, which is very close to what I wanted:
But much more often, I'll end up with an image of a freakishly proportioned plastic doll, a kiddo, or some equally undesirable and horrid combination of the two. What am I missing to get consistent, desirable results? If it makes a difference, the model I've found to give me what I asked for the most often is Flux Realisim with maxed out prompt adherence, but even that only works one time in ten.
k up, you need a computer."