I have yet to find a NSFW photo model that does natural pubic hair well. Some won't do it at all, no matter how many parenthesis you throw at it. So I decided to be the change I wanted to see in the world and trained this LORA to add full, glorious muffs to my gens.
Trained on ~100 images of beautiful beavers with Realistic Vision 1.3 as the base. All solo nudes, and nearly all basic frontal or "spread eagle" shots. It struggles with creative/contortionist poses, and I suspect it will fall on its face with hardcore. But for basic pinup stuff, it'll keep you drowning in vagitation.
Go gently, this is powerful magic! 0.2-0.4 weight seems to work best. Anything over 0.5 gets real squirrely. This LORA also has a side effect of occasionally making the subject smile more. No idea why. I guess the machine overlords can't yet tell the difference between horizontal and vertical smiles.
Stably Diffused 2.6, with and without the LORA at 0.3:
Hassan Blend 1.5 with and without the LORA at 0.2
URPM 1.2 with and without the LORA at 0.3:
Description
Initial (and likely only) version.
FAQ
Comments (11)
Looking forward to trying this one out. you might add tags: pubes, hair, bush
awesome! but how do you get this to match hair color? i put blonde in the prompt, but i still get dark bush.
good lower half but really bad at cropping out heads!
Could you please send me the model files which I can use for further training in DreamBooth?
If you don't mind, please reach me out in Telegram @kopyl or email – [email protected]
I get an error: TypeError: Got unsupported ScalarType BFloat16
Any ideas why? Using webui with --prescision full
Any thoughts of doing an update with regional-type pubes? Asian pubes, Indian pubes, Caucasian pubes etc ?
Thanks for making this!
Holy Hell, this is amazing! Merci beaucoup!
This makes the best pubic hair out of all the bushy bushes I've tried :)
This is the best bush Lora on Civitai, IMHO.
But has anyone been able to get ginger bushes? I can't seem to get them to match the hair colour no matter how hard I try :'( lol
One of my favourite Loras - it works so well in SD1.5. Now that I'm starting with SDXL1.0 I'm having a really hard time getting the bush right - please my I humbly request and SDXL version??