>>69660865I've been training on Pony haven't tried on animagine yet but I am doing basically the same thing I was doing with 1.5 in terms of dataset prep. Autotagging with booru tags and training I mean. So I think they should be the same in terms of training.
Also one thing I noticed accidentally is Pony LoRAs do work on animagine but a bit weaker. (not the other way around though).