触发词：<code>meitantei style,</code>如果觉得还原力度不够则可以尝试加入<code>meitantei precure,</code>这个 Lora 基于 60 张动画截图训练而成。打标打得不是很好，忘记关闭 character 和 copyright 的标签导致打标文件中混入了不少角色的标签。再加上这些动画截图仅仅选自动画前几集，导致容易出现过拟合现象（使用该 Lora 时，不详细描述服饰可能会导致角色穿上明智安娜的背带裙，同时也可能使得一些人物身上呈现明智安娜的人物特征）推荐降权使用，并与其他 Lora 或画师词组合成新的画风。这个 Lora 输入<code>akechi anna</code>也可以还原明智安娜的特征，不过体态特征会有所偏差。<hr />Trigger words: <code>meitantei style,</code>If you feel the restoration strength is not enough, you can try adding <code>meitantei precure,</code>This LoRA was trained on 60 anime screenshots.The tagging was not done very well. I forgot to disable the character and copyright tags, so quite a few character tags ended up mixed into the caption files. In addition, these anime screenshots were selected only from the first few episodes, which makes overfitting easy to occur (when using this LoRA, if clothing is not described in detail, the character may end up wearing Akechi Anna's pinafore dress, and some characters may also show Akechi Anna's character features). It is recommended to use it at a reduced weight and combine it with other LoRAs or artist tags to create a new art style.This LoRA can also reproduce Akechi Anna's features when <code>akechi anna</code> is input, but the body/posture features will deviate somewhat.

Files with SHA256: c1197b2fab7681afb745e8f0ce70052324a3f23a1723a218c963e6032d4a2e6e