YunYing | 云缨 LoRA版本v1.1CE (ID: 56847)

Yunying LoRA V1.1

CG like -----> V1.1

Cosplay like -----> V1.1ce

Yunying LoRA with a little updates, now can trigger more silver outfit looks and her childhood looks.

嗨呀，不愧是我！云缨小姐姐小更新啦！新增触发【赤焰之缨】皮肤和【小时候】皮肤,（禁止炼铜）

Maybe better than V0 or not

或许更好，或许不够V0， V0貌似也能触发

Adjust a bit training images, base model is

调整了一下训练集，底模是 -------------> meinaV9

VAE Using:

VAE我用 kl-f8-anime2 VAE

weight around 0.5-0.9.

权重0.5-0.9.

works well with 2.5D or Anime base model

和2.5D或动漫类模型相性好

Recommended prompt along side the LoRA (original outfit):

跟LoRA一起用(原皮):

pony tail, golden hair accessory, red and green choker, red and white outfit

===========================================================

• If you want to trigger silver skin use:

  想要触发赤焰之缨皮肤输入:

  yunying in silver outfit, red pants, silver hair accessories,

• If you want to trigger childhood skin use:

  想要触发小时候输入:

  yunying in her childhood, child, cute,

===========================================================

真人相性还是不太好，依然要多加佐料多揉搓

Still not good at generating realistic one, need more LoRA to adjust.

带云缨周游世界吧! : D

Take her around the world :D

一起来创作和分享云缨更多的美图吧！

Let's have fun and share her nice pics!

描述:

真人版云缨的LORA试验型

An experimental LoRA for realistic version Yunying.

50张COSPLAY图训练

Trained by 50 images of yunying's cosplay.

底模是 ----------> ChilloutMix

Base model is ---------↑

权重 0.5-0.9

Weight aroung 0.5-0.9

适合真人类模型

Works with real-person likeness model.

当前仅支持红白外套原皮版

Only red and white outfit version supported.

比CG版（V1.1）能更好一丁点出真人类型的图.

Works better than V1.1 in generating real-person like images.

如果你对此试验型LoRA有任何想法或者建议，欢迎在评论区留言，感谢

If you have a better way to create realistic-looking LoRA or any suggestions on how to improve this LoRA training, please feel free to comment below.

(真人版COSPLAY的原始素材有很多镜像，导致红白间色衣服在学习时候很可能相互影响，导致衣服的左红右白有时候很跳，下一步会优化原始素材镜像并改进这个问题。）

(The original images for the COSPLAY have many mirrors, which may cause the red and white colors of the clothes to interfere with each other during the learning process, resulting in the left-red-right-white pattern sometimes being inconsistent. The next step is to optimize the mirrored materials and improve this issue.)

训练词语: