Hey there!
I'm experimenting a bit with pony. I'm still a bit unsure of how this model works but it seems to show better results overall. It also allows for some crazy LoRAs that didn't work that well before. This is one of many attempts to make it work, please share your results as well, I'm very curious of how the prompting works since I'm still learning. I would love to see can you can come up with.
Recommended settings:
Unsure, yet. I'm using AutismMix checkpoint as of now with euler ancestral, 28 steps and sdxl vae.
Weights for the model are weird as well, but seems like anything from 0.6 - 1.0