This is a Lora model for Lo style dresses, especially when Loxxxx is commonly set as a sensitive word so that the fashion cannot be entered. This can be used to compensate.
"Lo fashion is a rococo style women's dress, mostly with gorgeous ruffles and complex elements, mainly divided into three types: sweet, classic, and gothic.".
usage method:
Introduce keyword: lo fashion plus a type, such as lo sweet.
Cover example:
1girl, full body, standing, smile, spring, lo fashion, lo sweet, pink hair, cowboy shot, pink eyes, petals, flower in eye, hair flower, detailed face, kawaii, cute, young, lake,
bamboo forest, clear sky, outdoors, one eye closed, open mouth, <lora:lo_fashions_v2:0.6>, <lora:Moxin_1010:0.3>
Negative prompt: EasyNegative, cowboy,
Steps: 32, Sampler: DPM++ 2M Karras, CFG scale: 7.5, Seed: 3239382580, Size: 576x832, Model hash: d1facd9a2b, ENSD: 31337
I also post a picture without this LORA as same seed for reviewing diff.
All examples seed is 3239382580 in Anything V3.
Upgrade for more training data, and reduce the prompt of the training set to make it easier to trigger.
Training 20 epochs, 15+5 for last 5 eps boomed in "no left space on disk", and do it again.
1. Права на репостнутые модели принадлежат оригинальным создателям.
2. Для подтверждения авторства модели создателям следует обратиться к сотрудникам SeaArt AI через официальные каналы. Нажмите, чтобы подтвердить
