This Lora is trained on the Pony Diffusion V6 Model. It probably only works well for models descendent from PonyXL (e.g. AutismMix)

Use Euler A sampler and Hires fix without any negative tags for best results.