Sign In

Aegipan (Tokyo AfterSchool Summoners)

0

Download

1 variant available

SafeTensor

109.14 MB

Verified:

Type
LoRA
Stats

16

0

Reviews

No reviews yet

Published

Sep 6, 2026

Base Model

Illustrious

Training
Steps: 1,680
Epochs: 30
Usage Tips
Clip Skip: 2
Hash
AutoV2
E87666896F
Trigger Words
aeg1pan_tas

Created on Civitai

Comfy_00996_.png

Aegipan from Tokyo Afterschool Summoners

  • This is an experimental model and it's also my first time to create a LoRA. There are still many problems on this LoRA, and it's unstable, so don't give too much hope on it.

  • 这是我第一次制作LoRA,模型本身也只是实验性的试水,模型本身存在许多问题,且非常不稳定,所以恳请各位使用时不要太期待出图效果。

Prompts

Due to my personal ability problems, I can't ensure that every outfit could be reproduced.

出于个人能力,我不能保证每套服装都可以被复现出来。

I will list most of the prompts, add or delete prompts as you need.

我会列举大部分构成需要的Prompt,请根据需要自行增添或删减。

  • Character Base Prompt / 角色基础Prompt

aeg1pan_tas, shounen, goat, curved horns, 
white fur, white hair, light brown hair tip, short hair, bangs, 
light brown eyebrows, yellow eyes, horizontal pupils, 
light brown hands, light brown toes, 
belly tattoo, 
  • Performance Outfit / 演出服装

crop tank top, sleeveless, patterned tank top, 
green sash, gold sequins, black underwear, thigh strap, sandals, 
  • Modern Outfit / 时尚服装

white sweater, sleeveless sweater, crop sweater, black sheer crop shirt, layered shirt, 
belt, black pants, black platform sandals, 
  • Student Outfit / 校服

Japanese school uniform, black jacket, open jacket, exposed belly, 
belt, black pants, one pant leg rolled up, white sneakers, 
  • Wafuku / 和服 (performances bad 表现不佳)

wafuku, black haori, haorigimo, hakama, tabi, geta, 
  • Madness / 癫狂状态 (This need replace some of base prompts 需要替换部分基础Prompt)

madness, orange eyes, glowing eyes, 
maniacal smile, mouth open, 

Current Known Problems

  • Due to limited and highly repetitive training data, the character's face tends to be overfit. Character in generated images will always showing his left side face to the viewer.

  • 由于训练素材的缺乏且高度重复,角色的面部会出现过拟合现象。角色会高频率以左脸面对镜头且微微颔首。

  • Details would be lost pairing to other art style LoRAs, but this needs something more to verify.

  • 在搭配其他美术风格的LoRA使用时,细节可能会出问题,但这一点需要更多验证。

  • The curved horns of character have a chance performances awful, sometimes even corrupted.

  • 角色的卷曲山羊角出现模糊、崩坏问题的概率较大。