如果您需要定制lora，你可以通过QQ：3692722976添加我You can support my work through <a target="_blank" rel="ugc" href="https://ko-fi.com/wiehhg9527">ko-fi</a> or <a target="_blank" rel="ugc" href="https://www.patreon.com/c/wiehhg">patreon</a>, thank you→ (´｡• ω •｡`) ♡----------------------&gt;<h3 id="v1.1">V1.1</h3>This character is Hysilens, from &lt;Honkai Star Rail&gt;(v1.1).This LoRA was trained using 3D models, significantly improving character accuracy while resolving v1.0's style overfitting and headwear color inaccuracies. It also enables a transparent abdomen effect (ribcage visible through the skin). This version is tentatively finalized, but may be updated if better training data becomes available. While trigger words can activate the character's core features, consider adding these tags to mitigate color interference and ensure accuracy:这次的lora是《崩铁》的角色-海瑟音，v1.1版本。本次lora使用了3D模型进行训练，角色还原度有很大提升，同时解决了1.0画风过拟合、头饰颜色不准确的问题，同时此模型可以生成出透明腹部的效果（透过腹部看到肋骨）此版本暂定为最终版本，如果有更好的训练集，会再训练集一个版本。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>blue eyes,purple eyes,black hair,purple hair,</code></pre>Clothes（服装）:<pre><code>red glove,see-through,black thighhighs,torn thighhighs,single earring,ribs,strapless dress,purple dress,long sleeves,hair ornament,blue dress,black gloves,half gloves,</code></pre>----------------------&gt;<h3 id="v1.0">V1.0</h3>The training dataset is sourced from official models, achieving over 1.0 fidelity accuracy. Should higher-quality training data become available in the future, an upgraded version will be released.For specific tags, please refer to the reference images I've provided.此lora的训练集有不少素材来源于官方模型，还原度高于1.0，如果后续还有更高质量的训练集，还会再出一版，具体tag请参考我展示的图片。------------------&gt;Follow me for more models—thanks! ♪(´▽｀)关注我，我会更新更多最新模型，谢谢！♪(´▽｀)

【Honkai Star Rail】Hysilens丨海瑟音丨セイレンス

海瑟音 v1.1-Illustrious.safetensors

Illustrious&noob_1.1

Illustrious&noob_1.0

这次的lora是《崩铁》的角色-海瑟音，v1.1版本。本次lora使用了3D模型进行训练，角色还原度有很大提升，同时解决了1.0画风过拟合、头饰颜色不准确的问题，同时此模型可以生成出透明腹部的效果（透过腹部看到肋骨）此版本暂定为最终版本，如果有更好的训练集，会再训练集一个版本。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 style="text-align: start;"><code>blue eyes,purple eyes,black hair,purple hair,</code></pre> Clothes（服装）:<pre style="text-align: start;"><code>red glove,see-through,black thighhighs,torn thighhighs,single earring,ribs,strapless dress,purple dress,long sleeves,hair ornament,blue dress,black gloves,half gloves,</code></pre>

Files with SHA256: cef66af6bbc341749cd882de48e31123bd550bd51bfcbb69a8ee6cdd318eb3b0