v2 = Larger dataset, more steps. Seems to require more prompting to get two faces in a shot compared to v1. Putting "mature male" in the negatives can help if you want the faces to look less detailed/wrinkled and more anime-like.