This model has only been tested I2V, and is definitely quirky. I don't expect it to work well with anything other than I2V. It's been trained on videos where the woman's face is facing the camera and a penis appears from the side of frame, and inserted into her mouth. Most of them are upper body or face only i.e no full body. They were also only 256px. Will train higher resolution and/or T2V if this gets enough attention.
If her mouth is far away from the edge of frame, and it is not realistic for a male crotch to enter the scene easily, the model will get creative and do weird things like make the penis 2 ft long. So make sure that the face is near edge of frame.
Prompt: A penis appears from the left/right/out of frame. He inserts his penis into her mouth. A man is moving his penis in and out of a woman's mouth. He is holding her head. Her cheeks are moving with the movements of the penis. After a second, he pushes his penis deep into her mouth, taking a moment to bulge her cheek out.
Disclaimer
This LoRA model is provided “as is,” for lawful and responsible use only. I make no representations or warranties of any kind, express or implied. I do not endorse or condone any illegal, harmful, or harassing use of this model and cannot be held liable for how others choose to use it. By downloading or using this model, you agree that all use is solely at your own risk and responsibility.
Description
FAQ
Comments (36)
Chipmunking 🐿️
Yes lol. I thought of naming it that, but it's also got this "face fuck" aspect. So made a compromise
VERY good!!!
Love it!!!
This is nearly a masterpiece. One of the best lora's on here. It's incredibly adaptable. The actual penis does not look super realistic in all situations (not the size, just the lack the detail), but otherwise incredible.
Yeah, will have to do a 512px run
@rollpollution193 that would be awesome.
The way these women just keep working is a tribute to the fighting spirit of today's medical personnel, I kid. Seriously though, while I prefer T2V because it's easier to make vids, this one looks fun enough I'll have to bang out some images. Thanks for making this. Just when I think, 'well there's no new sex stuff, I'm surprised'.
Yeah. Intended it to be a "free use" kind of situation :D
Thanks for the great Lora!
I've already thought about trying to do this myself, otherwise the rest are monotonous, need more of this, variable, that can be combined with other lora.
Remember to brush twice daily!
Observations after some more time with this. Higher resolutions tends to double things, so you'll get lots of double penis coming from two directions, or two penis coming from the same person. Also, higher resolution does not fix the lack of penis detail.
That being said setting the config to a higher value in the high noise version can result in some more intense results without too much distortion.
Yeah. I trained this on 256px for v1, so will probably train again on higher res
90% of the time I had one guy come in perfectly and then the second guy comes in from the left or a vag is generated i think because the AIO nsfw model has baked in lora's messing with it need to retry on non nsfw version of wan maybe that's the same problem
@allyourbase64dev537 Yes have only tested it on vanilla fp8 I2V. The other NSFW loras overpower this one.
AMAZING, thanks for making it!
I have downloaded the model, now not to run it ?? Can anyone plz help
That's... really too much to try and explain if you've gone at this in such a reverse order of downloading a Lora first...
But I suggest watching some tutorials on youtube on how to install Wan 2.2 locally, or how to use Wan 2.2 in other means if you can't install locally.
A newbie friendly option is to go for Pinokio Browser that can offer a "one click install" kinda thing for the UI, but you still need to look for the settings, get the Lora's etc...
Here's one video of that (Wan 2.1, but you can also use Wan 2.2 in that... I'm only just now trying to get 2.2 working well after nailing 2.1 and 2.2 is a bit more complex it seems...)
https://www.youtube.com/watch?v=eulqBmpw72w
@sarcasm83334 Yes, second that. I would suggest learning about ComfyUI and how to run things there first. Start simple. Maybe even just SDXL, which are image generation models. Learn how to use loras against that.
Then move into the video models. If you don't have local hardware, look for tutorials on runpod where you can rent GPUs. They also have templates where everything is preinstalled and you can just learn the software, as opposed to learning how to install/download everything as well.
This is the template I use:
https://console.runpod.io/hub/template/one-click-comfyui-wan2-1-wan-2-2-cuda-12-8?id=758dsjwiqz
double click on lol
@rollpollution193 Thanks. I have actually used ComfyUI for SDXL and Pony image gens, but somehow have felt lazy to figure out how to set up the modules and workflows for video.
I'll look into it, as I do think it'd give me much more control over tweaks. Thanks for the template, much appreciated!
Should I change any settings from the default settings of WAN 2.2 I to V 14B Workflow that comes with comfyui. I changed the two loras with those that you provided, and adjusted the resolution to 256, but my output videos look nothing like yours (a lot more blurry) the movement itself is working correctly though.
I usually generate at 512 or 1024. I have not tested higher than that. If you download and drag and drop the example videos, they already have a ComfyUI workflow embedded in them. So you can just see and use what I'm using. There are also notes there to download the exact versions of the models I'm using.
"If you download and drag and drop the example videos, they already have a ComfyUI workflow embedded in them." Oh wow i didn't knew this was a thing, thank you!
Godly. A-Tier.
can you make a lora like this for wan 2.1, my pc is not that good, i can only use wan 2.1 that can generate 5 second i2v video every 2 hour. this lora looks good as it preserve the woman's face, can you help me please
Sorry I don't think I'm going back to 2.1, the prompt adherence is too good with 2.2. I think you can still use only the low noise WAN 2.2 model with your existing 2.1 workflow and should work with this lora. It will take the same amount of vram as 2.1
@rollpollution193 ok thanks, anyway im a complete beginer, what is "workflows?" should i put it in a folder like a lora? sorry noob question but i hope you answer please, im just using pinokio with wan 2.1, and i dont know anything about the intimidating nodes
@choydolf Oh, I'm referring to ComfyUI. The fun thing about it is that a lot of creators use it, so if you like a video or image, you can simply drag and drop it to your ComfyUI and it will give you the "workflow" that was used to create it. That's what a workflow is basically.
I'd say try some simple tutorials on ComfyUI. It may be slightly more complicated, but it's less of a hassle in the long run. You can rent GPUs with ComfyUI templates built-in on runpod or similar websites, so I don't even have to learn to install or download anything. You can start from there.
@rollpollution193 where would i drag and drop an image/video in comfyui for it to learn the workflow?
@sherman785420 Anywhere in the ComfyUI browser window.
Does anyone know how to get it to make the man hold her head with both hands instead of holding his own dick? otherwise its still good
Well, i tried this lora only on closeup head pictures (literally just head on picture), and man always puts one hand on head, second hand out of frame, so he doesn't hold dick. Strengths 1/1, models - smooth mix wan2.2 i2v (don't tried on default wan2.2 i2v), prompt from description. Tried to add there both hands, but man do not want.
After months of using WAN 2.1 with Fusion x I still haven't understood one thing: the Lora Lows for 2.2 all work well, can I also use the "highs" together with the lows like for 2.2 or do I have to use only the lows with WAN 2.1?
For some reason sometimes I get the guy putting his dick into the womans breasts instead of her mouth. No idea why it does this sometimes. Anyone had the same problem/know how to fix?
please make this LoRa for LTX2.3
Details
Files
FF-v3-low-120.safetensors
Mirrors
66_FF_low_120.safetensors
FF-v3-low-120.safetensors
ffv3_l.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
Cheek-Low.safetensors
CheekFuck_LOW.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120 (1).safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
脸颊kouL.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
脸颊kouL.safetensors
Cheek-Low.safetensors
CheekFuckInsertion-low-120.safetensors
CheekFuckInsertion-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
wan2.2-i2v-low-oralcheekfuck.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
FF-v3-low-120.safetensors
LORA_I2V_Cheek_Fuck_Insertion_L.safetensors
test-28-low.safetensors
FF-v3-low-120 (1).safetensors
FF-v3-low-120.safetensors
CheekFuck_LOW.safetensors
Cheek fuck insertion I2V LOW - Wan2.2 - A penis appears from the left of frame.He inserts his penis into her mouth.A man is moving his penis in and out of a woman's mouth.safetensors
Cheek fuck insertion I2V LOW - Wan2.2 - A penis appears from the left of frame.He inserts his penis into her mouth.A man is moving his penis in and out of a woman's mouth.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.