Very sick of the continuous concept blend of SD 1.5, I've created this embedding model (the first of a series, which I want to call "NStyles"), to obtain exactly the clothing I wanted, a now I want to share my work with the community.Description:Embedding of 32 vectors, trained on a dataset of 54 images, to generate a faux leather dress with black see-through pantyhose. The style of the dress and pantyhose can be personalized: if you want any seams or zipper to disappear from the dress, simply put the words in the negative prompt. Also cleavage is possible, and suspenders or high-neck version are too. If the personalization is too strict, simply reduce the weight of the embedding (like :0.9 or :0.8).Activation prompt: "dress and pantyhose"

[NStyles] Leather dress and black pantyhose

V2 changelog:- Expanded the dataset from 54 to 86 images- Training resolution upgraded from 512 to 768- Added rectangle figure body type- Added side view angle- Added rear view angle

Files with SHA256: a8cd1c3bcae0ed169c382e86ce9367c1222e47ca7ea902da2f7f9a0ed6e6d608