My version of Shigure Ui, trained with AnyLora, using 38 images of multiple poses and outfit, customly edited in photoshop, focuses on art accuracy, using 20 epoch 8 repeats, and 2160 steps. Use it at 0.7-0.8 weight, CFG/Guidance scale 7.5, Step 30. Support me at <a target="_blank" rel="ugc" href="https://www.buymeacoffee.com/madafada1991">https://www.buymeacoffee.com/madafada1991</a>

Shigure Ui/しぐれうい - (AnyLora SD 1.5)

V2. v1 is too weak to be released here. Sorry. I take good care of quality releases. Tags; I guess you guys can learn the tags from danbooru or the images example.

Files with SHA256: 3edba2fcc43758d2c118e04b21ee5dee752cad992a6eaa93c4e8363fac2f5e72