Hello there! It's simple model of a concept "clothed and naked". This lora combines relatively good with anime and realistic models. Keywords like "collage" or "comparison" are good, but trained keywords are: "drundr, clothed and naked". Works well with other loras. Training dataset had included vertical images, but in my experiments I can't reproduce any image. Example images include clothed/lingerie although original dataset mostly included clothed/naked images... I hope lingerie is not something that banned by civit. Training settings included in lora metadata (I hope you will not find a way to find me by them, lol).Looking forward for your generations. Goodluck, stay safe and warm!p.s. don't know why is clipskip set 2 for examples, lora was trained on "1", but it works just fine on both values.

Files with SHA256: e165c9f2309a5a6e766c20a040634b55b6b2062fea0645c4ff54483710200b62