Wow, this is freakishly smart. I just posted about how I figured out a way to get GPT to write a multiple-chapter novella, but I didn't know you could do something similar with images and have some stability/continuity in the look of a character.

One suggestion: it seems to me that you must have some clever prompt prefixes and suffixes to get "children's book"-style pictures. Why not use something similar on the first screen?

It feels strange to have users write an actual GPT prompt such as "Write me a story about..." when most of that prompt could be hidden away (and optimized) in the background. This way, Users could just enter whatever they have in mind (e.g. "Michael is a brave boy and encounters wolves, who he befriends" might be a real user's natural way of describing the story), rather than forcing them to think in prompts.

Congrats on your app!