On top of also doing mid shrinks, I also tried going for after shrink or shrunken.
While it kinda got the idea, there's still a few things it missed, such as the background, despite prompting for a large environment such as objects, it didn't really pick up on that and left them kinda bland.
There's also no training on accessories, such as hairbands, wrings, jewelry, stuff like that. So they would also shrink down too. It's an idea I would also like to play with at some point.
But the model is still in it's early stages and like I said, will be improved and worked on slowly in the background.
Anyway, I hope people enjoy!