This Lora was trained on Iconographic art of the Byzantine era. It was trained without telling the model these are paintings so it believes it's the actual world. No need for trigger words.