This is pretty worth to solve problem! I thought about it many times before when writing an article.
Surely that it is super early phase of development, but if possible, the ability for user to choose the generated art style would be valuable. Oh also, the generated image that contain people face is deformed (common limit of generative image model, if your product can address that, it will be ready to used in commercial, I think).
Thanks for sharing and keep it up!