September 16, 2025: Version 2 Released
The purpose of Version 2 was to reduce the unnecessarily large file size. I have succeeded in significantly reducing the file size.
Have you ever felt that the “sound effects” tag is not effective enough? Then I have created a LoRA to enhance the tag.
Known issues
1. Decreased letter accuracy
2. checkpoints with poorly rendered “sound effects” are not improved much by using this LoRA
Trigger Tag: sound effects
recommended weight
LoRA weight: 1
“sound effects” tag weight: 1 or more
e.g. -> (sound effects:1.5), <lora:sound_effects-000015:1>
Description
FAQ
Comments (3)
ケモノ系のモデルやLORAと一緒に使用すると英語テキストになる
モデルが効果音を描画する力を強化するだけのLoRAなので、ケモノのような海外が強いジャンルだと学習量の差で英文になってしまう可能性はありますね。このLoRA単体ではちょっと対応策が無いです。
This LoRA enhances the sound effects and makes them easier to read. It would be great to have some kind of LoRA that enhances things like “squiggle line speech” (or squiggle text, I’m not sure what it’s called), such as screaming during orgasm scenes. I see it a lot in doujin works, but it’s very hard to generate. Does anyone have recommendations for this?