To avoid overfitting, it is better to mix with other lora models and adjust the weight to about 0.75. It depends.Use inpainting with "only mask" in 0.4-0.7 denoising to fix face.Try to use short prompt to get better result!Use chillout model to get jett coser!Trained on anything-4.5.The tagger word is not necessary, but I recommend it.

Files with SHA256: eb5aa581bf84a78946676522ce1c5579e8c1e485724239671b29759313f00870