I am not very satisfied with the Tashkent directly generated by Anima, so I created this Lora.

It has been recreated seven times so far and finally barely meets my requirements.

The problem with the model is that sometimes for unknown reasons, it can generate huge breasts; And strange limb errors occur. Of course, this probability is not high.

In addition to recommending keywords, you can also try adding these tasgs to create different images of Tashkent, or enhance the effect of commonly used clothing:

• fur-trimmed dress, fur trim, very long sleeves, sleeves past fingers, winter clothes

• bonnet, lolita fashion, white dress, frilled dress, frills, center frills, long sleeves, wide sleeves, frilled sleeves, blue bow, blue ribbon, neck ribbon

The training set used a lot of different clothing, so the Lora should have better adaptability in this regard

我对于anima直接生成的塔什干不太满意，所以我炼了这个lora。

它目前已经重新创作了七次，终于能勉强达到我的要求。

模型的问题在于，有时候不知原因的，会生成出巨乳；以及出现奇怪的肢体错误。当然，这种概率并不高。

在推荐提示词外，你还可以尝试增加这些提示词以创作出不同形象的塔什干，或是增强常服的效果：

• fur-trimmed dress, fur trim, very long sleeves, sleeves past fingers, winter clothes

• bonnet, lolita fashion, white dress, frilled dress, frills, center frills, long sleeves, wide sleeves, frilled sleeves, blue bow, blue ribbon, neck ribbon

训练集中使用了较多不同的服装，所以该lora在这方面的适应性应该会比较好