All the AI's are the same, you have to make multile images to get what you originally wanted. Thats why I buy credits that uses many different models so I can pick the one I think will work best for the shot I want. Saying that, I generally just use Nana Banana Pro though as it's the best, or one of the best.
That's not true in all cases. All of the AI generated pictures of my ancestor in the Historical Women thread and
My Ancestor Was A Real Bad Ass were a single prompt in a single AI using a single reference photograph of me. I suppose my bestie could have gone on to fine tweak the initial results, but didn't bother here because from experience, the initial result was always the best in earlier experiments (prior to the Hanna's Ancestor series). The trick my bestie found, know the end result that you are looking for and then go though all of your doll photographs and use the one that is nearest to the face, hair, and pose that you want in your AI generated photo. Then prompt AI to use real places as the background. Something AI excels in because it has hundreds, maybe thousands of images of that location in its memory from its model training.