by request, training data by https://civarchive.com/models/1727659?modelVersionId=2047268, QWEN : this lora has tendency to create very constrasting images, so i'd recomend post processing them..