What you described made me think me about how Photoshop uses AI in the back-end of a lot of its functions - like to fill in gaps in backgrounds or removing objects. I think the workflow you described, or something similar, is going to be the norm very soon (if not already).
In any case, I appreciate sharing the insight and the end result speaks for itself. My small critique of the walking phases is really only because I found the description so evocative compared to the images.