Verified: 4 months ago
SafeTensor
Type | |
Stats | 1,472 17,274 |
Reviews | (109) |
Published | Sep 5, 2024 |
Base Model | |
Training | Steps: 2,898 Epochs: 30 |
Usage Tips | Clip Skip: 1 Strength: 0.7 |
Hash | AutoV2 05C9268EEC |
Created on Civitai
Trained with CIVIT AI Trainer on on 450+ high resolution sharp images of women opening mouth wide, sticking out tongue, showing uvula, displaying teeth, roof of mouth etc.
I find the results not as good as the SD 1.5 and Flux LORAs i made for same training set. But maybe you have a use for it..
Seem to works best with Pony Realism like checkpoints (such as https://civitai.com/models/153568/real-dream or https://civitai.com/models/372465/pony-realism) , follow their instructions for Sampler etc.
The most realistic samplers seem DPM2 a and DPM++ 3M SDE , that kind.
It needs a slightly higher CFG than I have seen in normal SD or Flux checkpoints, like 5 to 6,5