It work with everything it even works with drawing, you can improve them, makes them more or less realistics and animate them.
here is some test I ve done on the (in)famous Candy from fection classic "Thoughtless Vendetta 2"

You can see how the software extrapolate the details (especially on mouth/hairs and eyes) you obtain a more detailed and "textured" face than the original flash animated one. Then you can animate it.

By playing with the software you can push the extrapolation of the original caption a little further the get an even more "realistic look"

(but personaly I prefer the first one (more doll like than the second one fits better with what a Candy should be)
What do you think ?

and by the way... (happy 2022 to all)