<h2 id="heading-1606">中文</h2><p>一次就成功了的时雨绮罗，效果是真的不错捏。头部还原的很好，换衣服也很容易。可惜的不对称的丝袜，小辫子和经典屁股披风没有学到~触发词为"ShigureKira",添加"hair ribbon,ponytail,armored dress,single legwear,white boots,detached sleeves,fingerless gloves,white scarf"来获得更好的输出。虽然有点长哈，但还原的确实不错。推荐权重0.8，换衣服适当降低权重到0.6，保留触发词和头部相关提示词。训练素材为纯黑背景，稍微添加点背景描述词防止出现纯黑色背景。<span style="color:rgb(250, 82, 82)">如果她的衣领很牢固，请添加“collarbone”去除。</span></p><h2 id="heading-1607">English</h2><p>The ShigureKira model was completed in one go, and the effect is really good. The head is restored very well, and changing clothes is also very easy. It's a pity that the model didn't learn the asymmetrical stockings, braids and classic butt capes~ The trigger word is "ShigureKira", add "hair ribbon,ponytail,armored dress,single legwear,white boots,detached sleeves,fingerless gloves,white scarf" to for better output. It's a bit long, but it's really well done. The recommended weight is 0.8, and the weight is appropriately reduced to 0.6 when changing clothes, while retaining trigger words and head-related prompt words. The training material is a pure black background, and a little background descriptor is added to prevent a pure black background.<span style="color:rgb(250, 82, 82)">If her collar is hard,please add"collarbone" to remove it.</span></p><h1 id="heading-1608">不允许商业用途</h1><h1 id="heading-1609">NO COMMERCIAL USAGE</h1>

Files with SHA256: 340b38b1ad2571054fc17b873956d644a44839f720e1c2bdd8bcebcec4cedeaf