V1: Stumbled upon a large image set of Hooters waitresses and figured I use it as a learning experience. Based on 500 Captioned Images (Needs refinement) of the traditional Uniform (White Shirt, Orange Shorts). Triggers: T5shirt (produces typical white tshirt, sometimes adds sleeve detail, will clean up my data set.), L0ng (Long white tshirt, fairly consistent), T4nk (Not working as intended, will revisit for V2).
V2: Curated Dataset a little better. Added newer Black Uniforms. Updated Triggers and added new ones. Some work better than others. play with LoRa weight and trigger weight and they seem to behave better. I may have overtrained this model a tad, still learning. Hope you enjoy. 0range - Orange Shorts, D4rk - Black shorts, B - Black shirts, W-white shirts.
V3: Doubled the data set which should add some character variety and help with some clothing options. Example photos were whipped together using wildcards. Haven’t full tested CFG or strength but seems to do pretty well. Did not us Adetailer to touch up face.
Increased dataset, tags remain the same. Should have better grasp of some tags. Still learning.
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
