PixAI 멤버십
PixAI
PixAI
Generate
Sign in
홈
생성
모델
Studio
NEW
콘테스트
Mio.2 창작
툴박스
NEW
가이드

사용자 LoRA

XL

Sakura Laurel - Umamusume Pretty Deby

제작

  • AI 이미지 생성기
  • AI 애니메이션 생성기
  • 툴박스
  • 테마 생성기
  • LoRA 학습
  • Mio.2 에이전트
  • Studio

찾아보기

  • 인기 태그
  • 랭킹
  • 모델 마켓
  • 콘테스트
  • 뉴스

소개

  • 가이드
  • PixAI 블로그
  • Tsubaki.2
  • Mio 소개
  • 콘텐츠 규칙

요금 및 도움말

  • 멤버십
  • 크레딧 팩
  • 연락처

모바일 앱

  • 앱 스토어
  • 구글 플레이
© 2026 PixAI
  • 개인정보
  • 개인정보 처리방침
  • 저작권 정책
  • 특정상업거래법
Sakura Laurel - Umamusume Pretty Deby - AI Model cover image
업로드 날짜
2024. 3. 29. 오후 5:55
이용수
14.4k
리뷰
보통  (1)
트리거 단어

shobofuku, oni_costume, casual_clothes, chrismas_costume, tracen_uniform

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

추천 LoRA

Overdose [SDXL] - AI Model cover image

사용자 LoRA

Overdose [SDXL]
11
黒レン / Black Len - 月姫 / Tsukihime - AI Model cover image

사용자 LoRA

黒レン / Black Len - 月姫 / Tsukihime
7
サクラチヨノオー(ウマ娘) Sakura Chiyono O(Umamusume) - AI Model cover image

사용자 LoRA

サクラチヨノオー(ウマ娘) Sakura Chiyono O(Umamusume)
4
lilkyo - AI Model cover image

사용자 LoRA

lilkyo
9

설명

https://civitai.com/models/42504?modelVersionId=417052

更新基于Pony模型训练了一版人物lycoris,可适用于pony系模型。导言这是一个《Umamusume Pretty Deby》《闪耀优俊少女》的樱花桂冠角色LyCORIS。在使用中使用cut off插件(https://github.com/hnmr293/sd-webui-cutoff)防止元素污染会获得更好的效果。示例图使用了Viewermix模型进行生成。使用方法建议以0.75左右的权重使用。在打标过程中,为隔离元素污染,采用了每个卡面分开设关键词的方法,以下是各个立绘的关键词:私服:sakura_laurel_CA特雷森制服:sakura_laurel_UN胜负服:sakura_laure_SF(打标的时候打漏了一个字母l,但是就这样吧)其他非特殊立绘:sakura_laurel训练细节以NAI Final作为底模训练,使用了约50张图片作为训练集,将约一半图片作为回归训练集,每张图片重复训练10次,训练10个epoch,最终步数为1500步。以下为主要设置参数:network_dim=32network_alpha=32keep_tokens=8conv_dim=4conv_alpha=4lr="1e-4"unet_lr="1e-4"text_encoder_lr="1e-5"batch_size = 2局限性人物立绘中的不对称设计的还原难度非常大,基本上只能依靠随机抽卡和手动辅助达到一定的还原效果。在出示例图的过程中没注意到Additional Network插件失效,一些LyCORIS的层没有输出,可能对最终结果有所影响。IntroductionThis is a LyCORIS of Mejiro Ardan from "Umamusum Pretty Deby". The difference between other similar characters Lora is the addition of Neige Émeraude. Using the cut off plugin( https://github.com/hnmr293/sd-webui-cutoff )to prevent elemental pollution will achieve better results.Example images are generated by Viewermix model.UsageSuggest using a weight of around 0.75.During the marking process, in order to isolate element contamination, a method of setting keywords separately for each card surface was adopted. The following are the keywords for each vertical drawing:Casual clothes:sakura_laurel_CATeresan uniform:sakura_laurel_UNShobofuku:sakura_laure_SF(I missed the latter "l" when tagging)others:sakura_laurelTraining DetailsUsing NAI Final as the base model for training, approximately 50 images were used as the training set, and about half of the images were used as the regression training set. Each image was repeated 10 times for 10 epochs, with a final step count of 1500.The following are the main settings parameters:network_ dim=32network_ alpha=32keep_ tokens=8conv_ dim=4conv_ alpha=4lr="1e-4"unet_ lr="1e-4"text_ encoder_ lr="1e-5"batch_ size = 2LimitationsIt is very hard to draw the asymmetrical designs in character portraits. Maybe a lot of tries and manual assistance is a certain way to achieve good results.During the process of presenting the example image, I did not notice that the Additional Network plugin was invalid, and some LyCORIS layers did not output, which may have an impact on the final result.

댓글