Try size: 512 x 768
Weights: 0.9
Clip skip: 2
Pictures from majicmixRealistic_v6 (Not recommended to generate pictures on anime models, you should choose a tool lora trained on Anylora or Anything V5 etc. base models / 不推薦動漫生成圖片，您應該選擇在 Anylora 或 Anything V5 等基礎模型上訓練的工具型lora)
Note: Recommended to use ADetailer, repair face and eyes. If there are too many crowd you can adjust the confidence threshold to 0.8. Eyes, hair ornament and clothes are all trained together, so it is regarded as an Asian style lora rather than a tool lora. Although most of the material has a black background, a small amount of the material is nature and has been lightly learning. But it can be easily covered by the prompt. If it is difficult to cover, please increase the weight of the prompt, especially at night. / 建議使用ADetailer修復臉部和眼睛，假如路人太多，可以將置信閥值調至0.8。眼睛、髮飾跟衣服都是一起訓練的，所以視為一個亞洲風格lora，而非工具性Lora。雖然素材大多是黑背景，但是少量素材是大自然，也被輕度學習起來了。但是可以輕易被提詞覆蓋，如果覆蓋有困難請提高提詞權重，尤其是夜晚。
(為了紀念我跟Chilloutmix，以及Koreandolllikeness的相遇，所以就做了這個Lora，命名原本不是彩色隱形眼鏡/彩色瞳孔放大片，只是為了讓大家好理解才這樣命名。對我來說，曾經的生命低潮是它帶給我許多感動，我的記憶中對它有特別的情感。雖然畫它的機會不多了，但是我可以將當時珍藏的作品以另一個方式呈現)
