Art style of QuAn(泉酱, 也可以叫QuAn_, QuAn___ or __QuAn_ <s>什么下划线排列组合</s>).This model is trained base on Anything 4.5 model and 640X640 resolution.Trigger word <code>avatar</code> is not required, it can make face larger, 但更容易出现一些奇奇怪怪的背景元素。Common uses: <code>simple background</code> , <code>upper body</code> , <code>smile</code> , <code>open mouth</code> or <code>closed mouth</code><h3>Versions</h3><code>v2</code> includes two model, <code>quan_style_v2</code> and <code>quan_style_v2-avatar</code> .<code>quan_style_v2-avatar</code> is specialized for making avatars, but more likely to generate artist's signature in the corner.我去除了训练集中大部分的无用的元素，但是我保留了签名部分，我觉得保留了一部分，才知道你生成的是泉酱风格的头像<h3>Block Weights</h3>If you use <code>sd-webui-lora-block-weight</code> plugin, you can specify weights to affect results.<img src="https://imagecache.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f33b104a-2942-4c76-9cf8-1ebd38669a00/width=525/f33b104a-2942-4c76-9cf8-1ebd38669a00.jpeg" />The <code>style</code> preset is <code>1,0,0.5,0,0.5,1,1,1,1,1,1,0.5,0.5,0.5,0.5,0.5,0.5</code> , according to the block analyser.This model has not received a license reply yet. If you believe that this model infringes on your rights, please contact me and I will take it down.

QuAn Avatar Art Style LoRA

quan_style_v2.safetensors

v2 avatar

Training is not completely finished yet.<s>其实有v1版本的，那个拟合效果超好但被不小心被我覆盖了啊啊啊</s>

Files with SHA256: 45a5e95405d96606b2d1f02df697e884156d95b82aea3b25080e9243b4022db9