생성 시작
역사 기록
비공개 창작vip-icon
상세 정보
Lora를 사용한 Flux text2img
원작

Lora를 사용한 Flux text2img

5.7K
0
2.8K
2025-03-27 13:45:16 업데이트

SeaArt 官方模型加载器已经支持FLUX标签和筛选,所以之前对于怎么找flux checkpoint和lora的内容就不再赘述。 <브르>目前flux是支持多lora的,但为避免应用界面过于复杂,我仅开放了一个lora位,另外默认加载了FaeTastic Details和mjV6,权重均为1。 <브르>单图的算力消耗约30PT,2-4图则会降低单图消耗,最低能到19PT每张图。建议增加出图的batch size获得最高的性价比。 <브르>SeaArt's official model loader already supports FLUX tags and filtering, so I won't go into detail about how to find flux checkpoints and lora before. <브르>현재, flux supports multiple lora, but in order to avoid the application interface being too complicated, 我只开启了一个lora位, and the FaeTastic Details and mjV6 are loaded by default, 무게가 1인. <브르>The computing power consumption of a single image is about 30PT, and the consumption of a single image will be reduced by 2-4 images, and the minimum can be 19PT per image. It is recommended to increase the batch size of the plot to obtain the highest cost performance.

평점 및 리뷰

4.4 /5
0 개의 평점

충분한 평가나 댓글을 받지 못했습니다.

no-data
데이터 없음