Trained on ~100 images, most of them portraits, which does give it some bias. More commonly referred to as Slit Pupils, but I tried to avoid that term since I have had issues with it in the past. Works best with a focus on the face or highres images so that eye detail is high enough. Adding terms where characters have slit pupils, like horns, demon girl, cat ears, etc, does produce better results. I generally use it at 1.0 weight, but it may produce weird artifacts at times. Used grapefruitV4 and meinamix_meinaV10.