A commission! I'd like to thank my assistant, Factiis, for helping me collect the dataset for this model.

Version differences: If you're trying to get as close to the anime as possible, use the 'More anime-accurate' version. Some of the outfits had barely any screentime, so Factiis used Flux Kontext to make more images of some of these outfits. The 'More flexible' version contains these additional images, at the cost of the art style becoming more "generic" (but still good). I also made the dimm/alpha half the size, so it'll be 109 MB vs. 218 MB. Basically, use the Flexible one if you're less concerned about replicating the art style of the OVAs!

➼Obligatory self-promotion: consider supporting or commissioning me at https://www.fiverr.com/metalchromex/make-you-a-stable-diffusion-anime-character-model or https://www.patreon.com/MetalChromeX so I can continue making cool crap for you all to enjoy for free!