LoRa model trained on 30 images of Emilia Clarke's face. Let this run for 30-40 steps at 768x768 for good results. Works with realistic and semi-realistic/anime models.