Doggystyle possition.
Tested with a merge of AbyssOrangeMix2 with Anything V3
Tested with Deliberate for more realistic results, the model with higher dimension may be better for anime style, the one with lower dimensions for more realistic results.
Description
Model with 32 dimensions (wrong!!!), good with more realistic models
EDIT: looking again, the only difference is the network alpha, now at 32, I don't really understand why it works better