This training increased the training set, optimized the tag, and Tifa became more "flexible" and easier to use with other Lora.
1. The recommended weight is around 0.75
2. This version of tifa is easy to be naked. Please adjust the tag appropriately.
此次训练加大了训练集,优化了Tag,Tifa也变得更加"灵活",更容易和其他Lora搭配使用。
1.建议权重 0.75附近
2.这个版本tifa比较容易裸体,请适当调整tag。
About v1:
This is a Lora model about Tifa. It adopts the method of mixed training of real people and Tifa games. There are still many imperfections as a version for the time being. At the same time, I hope you can give full play to your creativity and provide further ideas for my creation.
Model related issues:
1. It is recommended to use Chilout Mix output
3. The triggering weight of Lora model is 0.9, and the effect is good
3. The model is still powerless to generate images of huge model size
4. The model is mainly good at facial modeling, and may match better with other models in terms of movement?
Thank you very much for preferring this model=)
关于V1:
这是一个关于Tifa的Lora模型,采用真人和Tifa游戏混合训练的方法,暂时作为一版本还有很多未完善的,同时我也很希望大家发挥自己创造力,提供我创作的进一步思路。
模型相关问题:
1.推荐使用ChilloutMix 输出
3.Lora模型触发权重0.9效果附近较好
3.模型对于生成巨大型号尺寸的图片暂时还很无能为力
4.模型主要擅长面部塑造,动作方面可能和其他模型搭配起来更好?
最后非常感谢大家喜欢这个模型 =)