Sexy lingerie set LoRA for SD 1.5
Clipskip: 2
CFG Scale: 5 - 9
You can change the variation of lingerie set appearance in your prompt with certain trigger words within the positive prompt and negative prompt. Here's example of how it's shoulde be used:
(full lingerie set: mean everything in the set will appear)
Positive: full lingerie set, pink floral lingerie, bra and panties, garterbelt, white stocking, thighhigh
Negative: N/A
(without garterbelt: garterbelt will not appear)
Positive: pink floral lingerie, bra and panties, white stocking, thighhigh
Negative: garterbelt
(without bra: the bra will not appear)
Positive: pink floral lingerie, panties, bare breast, garterbelt, white stocking, thighhigh
Negative: bra
#notice: "bare breast" should included to make the bra disappear otherwise it will keep appear
(without stocking and garterbelt: stocking and garterbelt will not appear in the image)
Positive: pink floral lingerie, bra and panties
Negative: white stocking, thighhigh
And plenty of variation at your preference
Check the image to see the example prompt, negative prompt will make the certain part not appear while the positive prompt will keep it appearing.
You can change the camera orientation to see the lingerie set in different angle with trigger words: camera shot from the front, camera shot from the side, camera shot from the back
++ ATTENTION: Please be aware the the lingerie may not appear beautiful at the initial generation. You must generate the image first then pick your favourist one, move the picture to img2img/inpaint tab to improve it.
Description
Verson 1.0
FAQ
Comments (3)
I noticed that tattoos of a rose or climbing grass may appear on the body. By the way, this can be seen in your previews.
Thank for the comment, the pictures that I used to train this LoRA were from the woman who have tattoo of a rose ( I guess ) while wearing this lingerie set and there are so many of it I can't remove from the dataset. You can try using "tattoo, ink" with bracket () in the negative prompt so it may disappear.
Btw, thank you very much for reviewing my LoRA, much appreciated!
















