Hi guys, i finally managed to update the Lora to a new well working version.

For this version i created a whole new dataset.

The generated sample pictures were all made with epiCRealism v5 but it was also tested on a few other models (realistic, semi realistic and anime) and it worked pretty good.

For the Lora weight i personally prefer 0.6-0.8 with a trigger prompt weight from 1.0-1.3 depending on the model.