Another model that I am pretty happy with. There has been a ppppu model for a while now but this is based off of his art. here is a repeat of my previous description, as it all still applies.

There are caveats.

Mainly, you need to have a pretty strong negative for it to look good. Start with this one, and you should be fine.

(painting by bad-artist:0.8), (painting by bad-artist-anime:0.6), (bad-image-9600:0.7) (NG_DeepNegative_V1_2T:0.7), easynegative, censored, bar censor, upside-down, (worst quality, low quality:1.4), (depth of field, blurry:1.2), (greyscale, monochrome:1.0), lowres, bad anatomy, bad hands, text, error

That requires some TIs, but it's worth it.

The other caveat is the model I use. Use Corneo's 7th Heaven Mix with orangemix.vae.pt

I also use clip skip 2 sometimes

Trained using deepbooru captions, (because I am lazy) so use underscores in prompt for best results. I alphabetize my prompts, but this is likely pointless. 297 images used.

I would recommend using Lykon's detail enhancer as well, because it works incredibly well.