Update (5-12-26): I trained a (much needed) V2. V2 has a tweaked dataset, different tagging and training settings, and more! V2 should outperform V1 and also be more flexible. While this LoRA was not specifically trained on it, komi shouko and komi-san wa komyushou desu may help with if you want to make anime screencap-like images; I've included examples with sh0uk0 (the tag I trained the LoRA on) and with komi shouko.
--
The Komi Can't Communicate manga ended today. I spent a large chunk of time yesterday on this intricate model of her anime design as both a fan and as one last way to celebrate an almost decade-long run!
➼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!








