breastInClass -> nudify XL.
Are you enjoying fine breasts and perverting the life work of science researchers? Consider donating a dollar to the cause. Anything is appreciated :)
tl;dr:
Nude female frontals. Shapely breasts with small nipples. Strips away layers of embedded censorship in models.
⬇️ read the instructions ⬇️
What it is
This is a LORA that I use in text2img to get better NUDE FRONTALS.
It alters the body shape, particularly the breasts.
It's best at weights under <1 with accompanying keywords.
The data set is a collection of fully nude adult women.
What it is not
The dataset is all naked women with no faces. Any effect on clothing, men, flat chests, or faces is unintended. Funny, but unintended.
While appreciated for the exposure - using this for fully clothed women defeats the purpose.
It only works when you include it in the prompt (saddened I must tell you this).
It does not bias towards asians. I just like generating them. There are no dark skinned women in the data set unfortunately, but I'm not averse to making a version for this.
Installation:
Download the file
Place it in models\Lora folder (webui)
Usage:
You can also add it to the prompt by clicking "lora" tab in automatic1111:
You can also type it into the prompt, but it will not work if you make a single typo.
ex <lora:nudify:0.5>
The XL model requires the XL base / XL base with VAE model. It doesn't seem to work with the refiner, or anything else.
Adjust the weights. 0.1 - 1.0
Tips:
This is a fairly strong LORA best used in conjunction with low weights and other keywords like "naked", "topless". Work to get your image close to NSFW first, then add this to push it over the line.
0.5weight is pretty good for most purposes.Very high weights can cause the face to crop out.
Adding "small breast" or "huge breast" keywords can add more size variation, keeping the shape.
Adding stubborn loras or conflicting terms (e.g. "shirt") to the prompt can burn the image.
Base SD models are stubborn and don't like to be made NSFW. Expect clothes, blurred nipples. Try different weights and add more keywords!
Description
Reduced network size to create a smaller file - results may be different
Same dataset as nudify xl.
FAQ
Comments (19)
even the demo image has an extremely weird flattened pelvis and strange looking chest...
i don't think XL is salvageable for horny. not your fault, the model itself is cucked to the core until we get an NAI-XL leak or something...
You're right, it's proven much harder to jailbreak. I had to significantly raise training steps just to get it to stop blurring nipples, which it still tries.
Check out "Better than words", and say SDXL is not salvageable again. Outside of the variety of LORAs, SDXL will surpass any single 1.5 checkpoint in a number of weeks.
@xnihil0zer0254 I'm not particularly impressed with that one... everything that comes out of it has a strange, bland, samey look to it. The closest thing I'm seeing to good horny material from XL is coming out of Envy's checkpoints, but even those are just uncannily "off" somehow from 1.5 stuff. Merely producing realistic depictions of nude humans in generic poses is not enough to make XL noteworthy, though I'm impressed even that is possible now.
I'd genuinely love for XL to be uncucked, but even the best "realism" models in 1.5 have plenty of NovelAI leak DNA in them - without an entirely retrained base model akin to NAI for XL, we can only do so much with finetunes.
@NeuralEros I genned my first nudes with Jugg and RealVis and yea they look weirdly uncanny, the skin detail on the legs is to die for, I absolutely love it, but the rest of the body just seems off somehow?
@xnihil0zer0254 This comment aged like the finest milk ever left on the dashboard of a car abandoned along a Texas highway in summer. Bravo.
@DeviantRava haha yea we were just too early in the SDXL era! what's your favourite horny model?
Another option for "lite" models, besides retraining on a smaller rank is "resize_lora.py" available from both kohya and bmaltais.
--dynamic_method=sv_fro --dynamic_param=0.95
should keep a great amount of the training with good resizing. It decides which keys in the lora are most important, and downsize the least important.
the tits are looking very fake (rounded silicon) from the sdxl previews.
Can anyone confirm?
would you believe the data set is all women with real breasts? genetics is just unfair to some people
looks pretty good to me. There are "skinnifier" loras if you need it.
Add onsite generation please 😟
hey this is the best tits lora i tried , but it would be perfect if there would also be more natural large big breast or small tits, it does look a bit the same, roundish but overall its good
thank ya. i think the best way to do that would be in separate versions so the lora could retain its accuracy without trying to fit all breast sizes. doable, still.
and re: "natural large", these models in the training set ARE natural btw, they're just genetically gifted lol.
good
works like a charm
how did you do it. I can't seem to prompt mine correctly. I'm using flux on comfyui
I tried this on img2img with <lora:nudify_xl:1> as prompt and nudify_xl.safetensors [d0d3f70139] as model but i just get a blue canvas as output :(
it's not an img2img lora
Details
Files
nudify_xl_lite.safetensors
Mirrors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
useless0_xl.safetensors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
nudify_xl_lite.safetensors
xl nudify_xl_lite.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.