Been tinkering with this for a while. Still trying to figure out what prompts work best, since it's a merge of a lot of models.
There's also the anime-style version of this model (more like 2.5D) mCQAni
I'll update the description if this goes anywhere, including the recipe.
I used vae-ft-mse-840000-ema-pruned.ckpt for all image gens.
Pruned Model fp16 has baked in VAE.
Might be a good idea to include sfw/nsfw respectively into prompt/negative prompt depending on what you're looking for.
(note, I'm running macOS 13.3, python: 3.10.10, torch: 2.1.0.dev20230326 as of posting this)
Update (Apr 14):
Seems to work best if you include:
"((masterpiece, best quality)), highly detailed, photorealistic, photo of" at the beginning of the prompt, good results with "((perfect face, detailed iris, realistic pupils)),(high detailed skin:0.9), (visible pores:0.3)," at the end.
However, you should get good results with simple prompts as well. Check example images, there's a variety of samplers, prompting styles, with and without hires fix.