This model has been trained on 100 favorites from my collection of LoRA training epoch sample images. In most cases every image was from a different session, so this should result in some barely predictable variety, as well as making it play well with other models. Have fun exploring it!