<h2 id="heading-317">版本已更新，请仔细阅读更新后的LoRA简介，欢迎大家的尝试，之后还会制作一个针对面部特化的版本以供大家给藿藿换装</h2><h1 id="heading-318">取消,已经有人做过特化版本了，那我就不做了</h1><p></p><h2 id="heading-319">The version has been updated. Please read the updated LoRA introduction carefully. Everyone is welcome to try it. Later, we will also create a specialized version for the face to allow everyone to change HuoHuo's outfit.</h2><h1 id="heading-320">Canceled, it seems that someone has already made it.</h1><h1 id="heading-321"></h1><p></p><p>由于素材有限，所以这个LoRA的效果并不是特别好</p><p>等到实装之后可能会重新做一下</p><p>（那也得我抽得到藿藿）</p><p>关键词为Huohuo, <span style="color:rgb(243, 244, 246)">long_hair, green_eyes, symbol-shaped pupils, white_pupils, black_jacket, sleeveless_jacket, ahoge, white_socks, loose_socks, long_sleeves, shirt, hair_ornament, grey_shorts, black_headwear, ears_down, fox_ears, fox_tail, animal_ears, green_tail,</span></p><p><span style="color:rgb(243, 244, 246)">极度推荐使用LoRA Block Weight进行分层控制，推荐预设为1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0</span></p><p><span style="color:rgb(243, 244, 246)">推荐权重为0.8左右</span></p><p>她身上的各种配饰都不能稳定生成，只能碰运气</p><p>因为官方立绘中的藿藿是露肩的，但是3d模型并没有，所以你生成的图像中可能会交替出现这两种情况</p><p>如果你发现她的狐狸耳朵竖起来了，可以尝试提高<span style="color:rgb(243, 244, 246)"> ears_down 的权重，或者删除fox_ears 和animal_ears</span></p><p><span style="color:rgb(243, 244, 246)">她的尾巴无法稳定生成</span></p><p><span style="color:rgb(243, 244, 246)">symbol-shaped pupils, white_pupils 可以最大程度地还原她的眼睛</span></p><p><span style="color:rgb(243, 244, 246)">我不是很清楚该怎么描述她的衣服，所以你可以自己尝试调整关键词以获得更好的效果</span></p><p><span style="color:rgb(243, 244, 246)">其实这个LoRA真的不是很好用</span></p><h2 id="heading-1012"><span style="color:rgb(243, 244, 246)">但是</span>藿藿<span style="color:rgb(243, 244, 246)">真的很可爱</span></h2><h3 id="heading-1013"><span style="color:rgb(243, 244, 246)">我真的很喜欢她的袜子</span></h3><p><span style="color:rgb(243, 244, 246)">（噫，合乎舟礼哉）</span></p><p></p><p>Actually, the reason I made most of the models myself is that I couldn't find them.</p><p>Although there seems to be an AI image of Huohuo on Pixiv, LoRA doesn't seem to be publicly available.</p><h2 id="heading-1014">Why is it that someone posted LoRA just a few days after Ruanmei's illustration was released, but there is still no Huohuo?</h2><p>Due to limited resources, the effect of this LoRA is not particularly good.</p><p>I might redo it once it is officially implemented.</p><p>(But I still need to draw Huohuo.)</p><p>The keywords are: long_hair, green_eyes, symbol-shaped pupils, white_pupils, black_jacket, sleeveless_jacket, ahoge, white_socks, loose_socks, long_sleeves, shirt, hair_ornament, grey_shorts, black_headwear, ears_down, fox_ears, fox_tail, animal_ears, green_tail.</p><p>I highly recommend using the LoRA Block Weight for layered control. I suggest using the preset values of 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0.</p><p>I recommend setting the weight to around 0.8.</p><p>The various accessories on her body cannot be consistently generated, so it's a matter of luck.</p><p>Because the official illustration of Huohuo shows her with bare shoulders, but the 3D model doesn't have that, so in the generated images, these two situations may alternate.</p><p>If you notice that her fox ears are raised, you can try increasing the weight of "ears_down" or delete "fox_ears" and "animal_ears".</p><p>Her tail cannot be consistently generated.</p><p>Symbol-shaped pupils and white pupils can best reproduce her eyes.</p><p>I'm not sure how to describe her clothes, so you can try adjusting the keywords yourself to get better results.</p><p>Actually, this LoRA is not very useful.</p><h2 id="heading-1015">But Huohuo is really cute.</h2><h3 id="heading-1016">I really like her socks.</h3>

Files with SHA256: feb5f4af7687e1d9b38df4e366386daf987c919f4fe6c3b13002ecab1794d2a8