This LoRA is trained with stock pictures of the bra brand Nessa.The current LoRA knowns these following models:<ul><li>clarisse</li><li>cleo</li><li>klara</li><li>kora</li><li>leticia</li><li>mamma</li><li>moly black</li><li>moly nude</li><li>naomi</li><li>rubin</li><li>selena black</li><li>selena red</li><li>selena white</li><li>tatuaz</li><li>tess</li></ul>strength could be around 0.4~0.8as all image in the dataset have head out of frame and very simple background, the LoRA tends to cut head and simplify background. strength higher than 0.8 is not recommended.

Nessa bra

Less faithful to the training dataset than the SDXL version, but still produces very good images.Trigger words are the same than sdxl version.

nessa_bra.txt

nessa-kr2_v1.safetensors

krea2 v1.0

pony v3.0

sdxl v3.0

pony v2.0

sdxl v2.0

pony v1.0

sdxl v1.0

Changed optimizer from prodigy to AdamWProbably the final version

nessa-pony-v3.safetensors

nessa-sdxl-v3.safetensors

Files with SHA256: ab80ae6b6145389699ecd95001c7c1b5912998df9a6cfb5b861da253cf5fa831