Do-sunno? / どーすんの?

某ゲーム内のポーズを再現するLoRAです

層別調整済みなので使用するモデルの画風で概ね再現できると思います

しかしリアル系モデルでは表情の再現性が著しく低下します

リアル系モデルの再現どーすんの？？？？

使用するプロンプト

one eye closed, solo, bangs, looking at viewer, upper body, open mouth, hand on chin

画像サイズ等詳細内容はPNGinfoを参照してください

translated by chatGPT

This is LoRA, which reproduces poses from a certain game.

It has been adjusted using sd-webui-lora-block-weight, and I believe it can generally replicate the style of the model you use.

However, for realistic models, the reproducibility of facial expressions significantly decreases.

How should we handle the reproducibility of realistic models???

Prompt to use:

one eye closed, solo, bangs, looking at viewer, upper body, open mouth, hand on chin

For detailed information such as image size, please refer to PNGinfo.