用戶LoRA

SD

Touhoku Kiritan - voiceroid

#VOICEROID

Touhoku Kiritan - voiceroid - AI Model cover image
上傳時間
2023年6月26日 晚上10:20
使用數
16.03k
評論
暫無評論
觸發詞

t_kiritan, brown short hair, japanese clothes, pleated skirt

許可權
  • 允許圖片生成與分享
  • 允許使用者下載此模型
  • 允許生成圖片用於商業用途

描述

model author link : https://civitai.com/models/27103/touhoku-kiritan Model Author : 18bain_aiart

Regarding the use of characters, please check the Tohoku Zunko official guidelines.

https://zunko.jp/guideline.html

Learned at ACertainty.ckpt. Recommended strength is about 0.4 to 0.7.

The reproducible costume is only the official standard costume.

The Kiritanpo cannot be reproduced. Or rather, even if it is learned, it is difficult to control with prompts, so it was removed from the learning material.

The following prompts are valid for character and costume reproduction.

t_kiritan, brown short hair, twintails, ahoge, red eyes, japanese clothes, pleated skirt, 11yo,

If you want to reproduce it down to your feet, try adding tabi and zouri.

You can substitute touhoku_kiritan for t_kiritan, and kimono for Japanese clothes.

If you want to use only the knife-shaped corner, you can reproduce it only with t_kiritan.

If you want to use only the costume, you can reproduce it with only Japanese clothes and pleated skirt.

If you use Japanese clothes (kimono) for the prompt, the background will tend to be a Japanese-style room, so adjust as necessary.

The horizontal composition seems to reduce the recall rate. Adjust as necessary.

Headgear was automatically attached by Tagger, but it is better not to use it at the prompt. The corners of the knife shape will be deformed.

If you get animal ears, try putting animal ears in the negative prompt.

See the author link above for details.

評論