사용자 LoRA

Dagasi Style, Artist Style LoRA for NoobAI/Illustrious by novafoxalt

#퍼리 #닌자

업로드 날짜: 2025. 4. 19. 오전 4:01
이용수: 7.9k
리뷰: 훌륭함 (3)
트리거 단어: dagstyle,
권한: 이미지 생성 및 공유 허용
다른 사용자가 내 모델을 다운로드하도록 허용합니다
생성된 이미지의 상업적 이용 허용

설명

https://civitai.com/models/1431648?modelVersionId=1679596

Original notes, novafoxalt:

This model is focused on replicating Dagasi's style with maximum flexibility, but it also recognizes their OCs with good accuracy. Use e621 tags to prompt OCs.

Updates:v1: initial release.

v2: Increased lora size, increased training steps, and lowered learning rate. Appears to capture fine details better with improved anatomy. I've barely tested this model yet so I'm not certain of the magnitude of the improvements.

v3: Increased lora size, increased training steps, and increased learning rate. This version seems to have better flexibility and more consistent style, especially around the eyes. This version, however, has also captured the concept of spiral eyes, and tends to partially apply it to the OC characters. If this poses a problem during generation, try adding "[@_@]" to the negative prompt.

This model is not to be used for commercial purposes and generations should never be sold.