如果您需要定制lora，你可以通过QQ：3692722976添加我----------------------&gt;【Illustrious&amp;noob_1.1】This character is Jahoda, from &lt;Genshin Impact&gt;.Compared to previous versions, this training set utilizes 3D models for training, achieving higher accuracy. Theoretically, the character's appearance can be restored using trigger words alone, but color-related tags may affect character features. You can use the following tags to ensure character fidelity:这次的角色是《原神》的角色-雅珂达。相较于之前的版本，这次训练集使用了3D模型进行训练，有更高的准确度，理论上说，仅通过触发词可以还原人物形象，但是可能会因为颜色相关的tag影响人物特征，可以使用下方tag来保证人物还原度：Appearance(外貌)：<pre><code>orange eyes,braid ring,multicolored eyes,pentagram pupil,light blonde hair,twin braids,</code></pre>Clothes（服装）:<pre><code>scarf,black shorts,short shorts,black scarf,torn clothes,black belt,white clothes,hat,black headwear,sunglasses,eyewear on head,torn thighhigh,tinted eyewear,single thighhigh,black glove,fingerless glove,light grey thighhigh,mechanical hands,single glove,</code></pre>----------------------&gt;【Illustrious&amp;noob_1.0】This character is Jahoda, from &lt;Genshin Impact&gt;.Due to insufficient training data, this LoRA still has some limitations. An updated version will be released when higher-quality datasets become available.This LoRA activates character features via trigger words, but may be influenced by color-related tags. For optimal accuracy, consider adding these supplemental tags:这次的角色是《原神》的角色-雅珂达。目前因为训练集的数据不够，lora还存在一些问题，等后续有更好的训练集会再次更新。此lora仅用触发词就可以还原人物外观特征，但是可能会被和色彩相关的tag影响，所以可以考虑添加以下tag来保证人物的还原度：Appearance(外貌)：<pre><code>orange eyes,braid ring,multicolored eyes,pentagram pupil,light blonde hair,twin braids,</code></pre>Clothes（服装）:<pre><code>scarf,black shorts,short shorts,black scarf,torn clothes,black belt,white clothes,hat,black headwear,sunglasses,eyewear on head,torn thighhigh,tinted eyewear,single thighhigh,black glove,fingerless glove,light grey thighhigh,mechanical hands,single glove,</code></pre>----------------------&gt;Follow me for more models—thanks! ♪(´▽｀)关注我，我会更新更多最新模型，谢谢！♪(´▽｀)

Files with SHA256: 015dbc2e853b00de0b5bf6d7f4fb8c0792572fd4316a3f63aab0a3862d04be41