HIKA / Style / IllustriousXL

v2.0 update : no difference in datasets, but use text encoder for general use, and increase total learning steps

weight : 0.8 ~ 1.0

500 training images from pixiv HIKA

intended to produce mature contents

artist tag that I use : (artist:kedama_milk:0.85), (henreader:0.75), (donedone:0.8), (tianliang_duohe_fangdongye:0.85), (ciloranko:0.9), (artist:chen_bin:0.85), (ask_(askzy):0.8), (reel_(riru):0.85), (parsley-f:0.85), (shuri_(84k):0.8), (uekura_eku:0.8), (noyu_(noyu23386566):0.8), (yumenouchi_chiharu:0.9), (hyou_(pixiv3677917):0.9), (eumi_114:0.8), (chon_(chon33v):0.8), (maccha_(mochancc):0.9), (fuzichoco:0.9)