This version has less overfitting and gives better results.
This model is trained on photorealistic images of women. This works best with portraits.
Combine this with a photorealistic checkpoint model. This does not require any trigger words.
