So this is just antimage person without any skins, trained on 640 x 640
Added some nsfw and sfw arts to dataset, so lota is improving nowAlso fixed trigger word)
wei-10.safetensors