To generate the images, I recommend using ADetailer and EasyNegative.

I recommend using a VAE of your choice. For generating these images, I used vae-ft-mse.

Versions C, B, A, SS, and X will be available soon on my Patreon and Gumroad. Each version generates slightly different artwork. The D version (from this post) produces a more realistic style, but depending on the LoRA used, it might lean more towards anime.

Please do not use this model to generate CSAM content. Use it responsibly.