If generating shadows, is it enough to just provide the original image (without shadows) and the target image (with shadows added)? Or should the target person mask and the shadow mask also be added?

[gangxu822]