Quick Generate
Model Description by Creator
本LORA模型是由LLAS手游的立绘图训练的，共计1500+张图，训练了20epoch。所有训练图片均为横*纵：1024*768
This LORA model was trained by over 1500 pics of mobile game of LLAS, with 20 epoch. All of those trained images were 1024*768.
该LORA的效果主要体现在能让画面的饱和度明显增加，将非常多光效、特效的效果体现出来。
This LORA have effects on increasing the saturation of the picture and reflecting a lot of light effects and special effects.
并且因为LLAS内的服饰非常华丽，使用此LORA在绘制裙子、卫衣、礼服等众多服饰的时候会变得更加华丽。
In addition, because the clothing of LLAS is very gorgeous, using this LORA will become more gorgeous when drawing skirts, hoodies, gowns, etc.
使用此LORA不需要添加trigger word，您可以开比较小的权重来让画面产生不同的效果。
There is no need to add a trigger word with this LORA, and you can use a small weight to give different effects to the image.
由于本人刚开始尝试制作画风LORA，因此多少会存在一些问题。由于训练集全为横向的图，在纵向生成的时候可能会出现一些肢体错误（尤其体现在手）由于LLAS画风比较强烈，其中眼睛的变化尤为明显，因此如果不希望产生太大的变化，您可以降低权重来使用，或者不要用此LORA绘制面部。以上问题我将会持续研究改进，如果有新的问题发现请务必提出，感谢您的支持。
As I just started making painting style LORA, so there will be some problems. Since the training set is all horizontal drawings, there may be some body errors (especially hands) during the vertical generation. As LLAS painting style is strong, the changes of eyes are particularly obvious. Therefore, if you do not want to have too much change, you can reduce the weight to use it, or do not use this LORA to draw the face. I will continue to study and improve the above problems. If you find any new problems, please be sure to raise them. Thank you for your support.
国内模型站点：
