사용자 LoRA

XL
캐릭터

Galatea (Claymore) - SDXL

#클레이모어

Galatea (Claymore) - SDXL - AI Model cover image
업로드 날짜
2025. 8. 10. 오전 6:40
이용수
143
획득
12,288
리뷰
훌륭함  (1)
트리거 단어

Galatea_Claymore, long hair, blond hair, silver eyes, armor, pauldrons, full body leotard, gray leotard, gray leggings, cape, knee boots, armored boots, vambraces

권한
  • 이미지 생성 및 공유 허용
  • 다른 사용자가 내 모델을 다운로드하도록 허용합니다
  • 생성된 이미지의 상업적 이용 허용

설명

Trigger words: Galatea_Claymore, long hair, blond hair, gray eyes, armor, pauldrons, full body leotard, gray leotard, gray leggings, cape, knee boots, armored boots, vambraces

Recommended weight: 1.5 for her transformed outfit. I think 1.0-1.5 can still work for her face with a different outfit. Not tested below 1.0.

Please make sure you read the notice below:

Galatea is an important side character from the old manga/anime Claymore (クレイモア). She has two very different outfits. This LoRA ONLY presents the warrior outfit from the first half of the manga. Although she has a relatively higher number of scenes, which allowed me to capture more pictures than other side characters, the bad resolution and blurry face details make it difficult to generate her face details properly. The main problem is her eye color. In most scenes, the environment is very dark or significantly affected by sunlight (such as sunset). I set the prompt of "silver eyes" as the default, but it is okay to change it to "gray eyes"; it still works, though the result probably will not be much different. Please always turn on the "Face Fix" function, which will help a lot—at least the eye color will be much more stable. Her skirt-like armor often generates a wrong result. The number of them is six, but the results may be fewer. If you care about the accuracy, you need to try several times or cooperate with pictures to limit the output result. It's possible to generate her Claymore sword together; just add some prompts such as "sword" or "holding a sword. It's doable, but there is no guarantee the result will be as good as the original scene. You need to adjust several times to get the ideal result.

Overall, compared to the SD version LoRA I made before, this version has significantly improved; almost every detail can be displayed clearly.

댓글