For more LoRAs and updates, Join my Discord:
https://discord.gg/ZVWVhT43GW
I've been trying to train a pussy LoRA for more than a month and the amount of time I spent scraping a dataset and doing training runs will go to the grave with me.
Before you use this LoRA know that this LoRA is NOT perfect, but it's pretty good when it works.
Trigger words are varied and mostly revolves around the word "Vagina".
The word "pussy" was NOT used in captioning.
It has a fair understanding of pubic hair and shaved vaginas you could try to prompt for:
Thick dense pubic hair
Shaved vagina
Trimmed pubic hair
A small patch of pubic hair (rarely works)
It also has a decent understand of angles, you can try to prompt for:
shaved vagina from behind
vagina from above
vagina from below
Additionally, you could try to prompt for:
She is spreading her labia with her fingers.
I2V:
I did not test this on I2V as I almost never use I2V, I create my LoRAs for myself and my use cases and share when I'm fine with the output.
You're welcome to test and leave your thoughts in the comments.
LoRA was trained using 2xH200 running on parallel using my LoRA Training template on RunPod.
You can use it here:
https://get.runpod.io/diffusion-pipe-template
Or see a video about it here:
==================================================================
Created by HearmemanAI
Description
FAQ
Comments (59)
good shiet
Would it be possible to upload this as two different safetensors files instead of a zip? If it is a zip file addons like Comfyui Lora Manager can't automatically match to it and download metadata.
Be less lazy
@axicec Less lazy? Seriously? It's a valid request. Lora managers won't be able to archive metadata and preview images. Don't be dumb and post useless dribble.
@axicec - consume fecal matter, sub human troglodyte
@axicec Be less of a jerk.
@kamranahm880 Bro just double click on the zip file and extract the two safetensors to your lora folder. That's what he means by not being lazy.
@DigitalCandy - you are so dense you think you are the only person who knows how to unzip files? You look amazingly smart here. The issue is about scripts that use the civitAI API to retrieve metadata like triggers and thumbnails and such... if the creator uploads a zip, none of the software works. If the creator uploads the safetensors, all of the software works. It has absolutely nothing to do with being lazy.
Some of the stupidest people on the planet are the most arrogant. It would be hilarious if the world weren't so dysfunctional due to these same people.
its just sketchy downloading the zip when its the same effort to not upload it in the zip. I dont even know why Civit allows it
After you unziped the file right click on lora manager and choose relink to civitai and paste this model URL, should work and download the preview etc
@dxjaymz Oh thanks this is very helpful.
Finally some good fucking lora
I am testing this lora on I2V and the result is not as good as in the sample images. I am using Euler Simple and the latest Lightning Lora
Well it is a T2V lora. Sometimes settings will have to be different to get the same out've it. Generally you can use I2V base loras on T2V workflows, too, but again with different strengths. Also the samples say they have other loras applied too, so it would make sense to look at those too
Did you try reading the description? This is T2V only.
Awesome👍🏻. Can you do same for 5b?👍🏻
No, sorry.
and kindly where the f''+ is the workflow?
save the image and place in comfy. Sometimes creators dont share flows on all images, but generally the first few should
@llhappyll822 My workflows are not embedded in the videos.
@HearmemanAI my bad, I thought I had downloaded one of your samples. I tend to grab the first one as a reference, but yeah I never did in this case. Not too important when its flexible enough
Finally a good va-jay-jay. Thanks Hearmeman.
Glad you like it
i am looking for a i2v lora
I'm looking for a Maserati and a blowjob
I never got to experience the onset of eternal september, as I was one of those dweebs. But this is very similar I reckon
Be the change you want to see
would be nice l will pay you to train it
i2v工作流,模型权重在1.5-2表现不错,但是会改变人物的面部。
Hey Hearmeman, does your Diffusion Pipe template a good preset for SDXL training?
Tested it in I2V , sometimes it works , sometimes the anus is missing , something you get some eldritch horror , but most of the time it works just fine , what worked for me is making a more descriptive prompt of the vagina and anus , I'm specifically talking about I2V.
Anyways amazing work dude.
Updating , in I2V the the butthole rarely appears , the vagina appears 100% of the time.
@idostuffin3d Have you tried prompting for anus?
@HearmemanAI I did , tried many different ones , with many different descriptions ,repeating each one with different seeds , still the anus appeared rarely.
@HearmemanAI training with the i2v model the same dataset will produce much better results in i2v, just saying.. xD
Tested in i2v - same missing anuses.
@idostuffin3d have you find another way to make an anus appear in i2v other than this?? Seems like mission impossible here! haha
awesoooome, results are amazing :D
I hope you'll give us the same one for Qwen one day :3
lora is called "pussy", but word "pussy" is not in captions 🤔
This is because wan 2.2 understands vagina/vulva better than pussy, so I imagine that since it's a concept its already somewhat familiar with, using these words allows for more advanced learning.
@WearCatEars exactly
@WearCatEars now I've gotta try it, I've been getting peanut butter instead
the only one which can draw a good anus
and the i2v can you do plz
This is the best looking and most realistic pussy and anus LORA for WAN 2.2 I've tried. Great job! However, there seems to be very large bias towards shaved pubic hair. I rarely can ever get it to generate pubic hair regardless of my prompts, and when it does, it usually looks extremely bad. Perhaps adding more hairy pussies in your dataset could help.
Please PLEASE a dedicated i2v version of this!
made lora for pussy but pussy is not a trigger word!?
He created the vaginal lore. What's not clear?
FYI, this isn't showing up in the Power LoRa Loader node's Show Info window.
amazing! was able to run this with WAN 2.2 iTv with no problems using 12GB VRAM
how to make that get draw complete form and texture but not affect character skin?
Please have separate download links for high and low noise. ComfyUI Cloud will not allow this to be imported if it's in a zip file.
for some reason lora won't show up in possible picks after wan model is selected. like it's a checkpoint, not a lora. makes it hard to add to used resources.
Wow this LORA works amazing on I2V...you should continue perfecting it. I would certainly pay for it.
how do you get it to work on I2V? I haven't gotten decent results at all with I2V. Author said himself it was not trained for i2v
for me its working with i2v but not a perfect result yet and the a***hole is also not showing
Couldn't you have uploaded LoRa in SafeTensor format? Now how do I load it? Honestly, some users don't think like native developers. Still, the effort is appreciated.
bro u have both high and low inside a rar , i feel like u dont develop much
Take your head out of your ass.
"native developer" lol
Details
Files
PussyLoRA_Wan2.2_HearmemanAI.zip
Mirrors
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip
PussyLoRA_Wan2.2_HearmemanAI.zip