这是一个修复所绘制对象色彩的功能性LoRA。推荐日常使用权重为-3到3之间，您可以使用正权重来增强图像原本的色彩，如果您希望使用该lora生成类似的黑白图也可以使用负权重降低图像的色彩饱和度（相比于正权重可能会对构图产生影响）。本模型在生成竖图时效果最好。This is a functional LoRA for restoring the colors of drawn objects. For daily use, a weight between -3 and 3 is recommended. You can use a positive weight to enhance the original colors of the image. If you wish to use this LoRA to generate similar black-and-white images, you can use a negative weight to reduce the color saturation of the image (though compared to positive weights, it may have an impact on the composition). This model works best when generating vertical images.基于Kohaku-XL Zeta模型进行训练，不建议在其他模型上进行使用（即使会获得一定的效果），我也会抽时间制作该lora的其他模型版本Trained based on the Kohaku-XL Zeta model, it is not recommended to use this LoRA on other models (even though some effects may be achieved). I will take time to create other model versions of this LoRA.顺便说一句，作者个人认为此lora用1.2权重起手是一个不错的选择，权重低于1.2对于图像原本颜色的改变较为微弱，+2在大多数情况下应该可以满足您的需求By the way, the author personally believes that starting with a weight of 1.2 for this LoRA is a good choice. A weight lower than 1.2 will make relatively minor changes to the original colors of the image, while +2 should meet your needs in most cases.
