Damn, how can you sit so well?! 😮
I've been practicing 😏
Wanna support me?
Don't send Buzz, send me USDc or USDT through Polygon Network:
0x3f1e1323fa90b4c9f5a386bfa5b1e72f5a6c25e9$0.10 or $10, any amount is welcome and helps me make more LoRAs for the community
Workflow
I don't use ComfyUI, so I don't have a workflow to provide 🤷🏻♀️
The training is done on Wan 2.2 A14B Diffusers (BF16)
Training
400 steps on high noise
3 edited videos (80 frames each)
HLR + ZCD (These acronyms were created solely to confuse you)
Lots of hope
$10 (😭😭😭😭😭😭😭😭😭😭😭😭)
Usage (Low noise)
This LoRA was trained only on High Noise, which means you'll have to use some Low Noise that knows what a penis is (I used the Low Noise from POV Insertion V1).
Got an idea for LoRA? Then let's do it! (or at least try 🤠)
In my repository, I created a model called "LoRA Requests", so if you want to request a LoRA, send videos and photos to help build the dataset for a public LoRA that will be posted here, that's the place to go.
Open LoRA Requests on HuggingFace 🤗
I2V:
Prompt (ongoing):
At the bottom a man's lower body appears, his belly, thighs, hand, and his hard penis are visible. He is holding his hard penis with his left hand, stroking it.
With both hands, she separates her buttocks and sits on the hard penis, inserting it completely inside her pussy.Prompt (cut to):
Maybe not, bro, at least not here 🙂🤚🏻
Visit this LoRA on HuggingFace for uncensored examples, I know you want them 😏
High Noise LoRA Scale: 1.0 for creativity, 1.5 for stability or "cut to"
Low Noise LoRA Scale: 1.0
Shift: 4
T2V:
It probably works, but I haven't tested it. If you're going to test it, just keep the same structure as I2V, and try starting with a very low Scale on High Noise, like 0.5
About HLR + ZCD
This is a fast-learning technique, which makes LoRA less flexible. This can drastically reduce creativity, but it yields stable results while using few resources. Negative effects:
It will probably only work at the same camera angle
High chance of not responding to different prompts
High chance of forcing the original characters of the training video Training only the High Noise reduces this chance, but it can force things that are VERY similar, like the same hair length, same color, etc.
Download
Download High Noise LoRA (On HuggingFace)
Download Low Noise LoRA (low-pov-insertion-v1.0)
Help me creating more
If you want to help me continue making LoRAs, or if you want me to make a LoRA for you, buy 5000 PlayCoins at Aisha-AI and transfer them to my account (account number 2).
This helps Aisha-AI to stay alive and produce new LoRAs for you all 💜
Description
First version
FAQ
Comments (24)
Hello there. I'm curious, how much time does your training take for those high-noise only LoRAs? Thank you 🙌
A little over 3 hours
@AishaAI Thank you
Please create poses and movements that show the woman's face
I think you need to explain better
@AishaAI What I mean is that in this Lora animation, as well as the "Riding a bike dildo" Lora you posted before, the characters' faces are not shown, which defeats the purpose of "cut to," because any character would look pretty much the same from behind. I suggest adding an animation where the character turns to look at the camera.
By the way, based on my current testing, the facial consistency in your "BBC Reverse Squatting (front view)" Lora seems a bit off.
@jevencheung I don't need to add that, just write it in the prompt.
Want the person to look back? Write "Then she looks back to the camera" 🤷🏻♀️
You don't need to make a LoRA for something as simple as looking back, just write it, Wan knows how to do it.
Regarding facial consistency, you need to quote the person's face at the beginning, and then quote it again after the cut.
Example:
The video starts with a pretty-faced woman posing. Then the scene cuts to the same pretty-faced woman doing [etc, blah blah blah]This will make Wan look at the face in the initial image and use it as a reference, like "Ah, this is a pretty face..." and then you quote it again and he repeats "Ah, I need to put that pretty face."
If that still doesn't work, you can force the weight of the tokens:
The video starts with a pretty-faced woman posing. Then the scene cuts to the same (pretty-faced woman:1.2) doing [etc, blah blah blah]@AishaAI Thank you for your answer. I've been looking for a suitable solution for processing batches of images to address the issue of facial consistency, because the expressions and poses of the people in each image vary greatly. It seems I'll need to adjust the prompts for each image individually.
Wait 🙋🏻♂️ will this work with a "backdoor" insertion low LoRA?
Probably not, since I was trained with vaginal penetration
I have been trying to mix in other loras, but so far no luck. I think it would require a low noise "pov insertion" that is capable.
@cnt_montecrypto The only thing Low Noise does is provide details, textures, the smallest things. I don't need to create a Low Noise LoRA, you just need to use a Low Noise LoRA that already knows what a vagina is and what a penis is, and preferably one that also knows what vaginal penetration is
@AishaAI Yeah, what I meant is in order to handle "backdoor" as @TieFighterPilot mentioned, I think the low noise would have to understand that. Or are you saying that the High Noise also would need to be capable?
@cnt_montecrypto High Noise is responsible for movement and positioning, while low noise is responsible for finer details like texture and other micro-movements. Give me a video similar to this but with anal penetration and I'll make a LoRA
As if you can actually train a LoRa with only 3 videos
So...?
@AishaAI Not saying it's a bad thing bro
@OwnPear8996980 It seemed like an incomplete question
@AishaAI some people use "as if" in a similar way to how others use "wow"
@HarryPsalms Oh ok 🤣 I didn't know that
@HarryPsalms sounds like an ESL thing, lol. it's not how anyone speaks in the US. "as if" = "you can't"
Your output is ridiculously high, bro, but I like it, keep going!
Lol those insertions look completely off, as if entering the clit, a hit and and miss
awesome, working great and thanks for the low recomendation!
Details
Files
sitting_penis_high_noise.safetensors
Mirrors
sitting_penis_high_noise.safetensors
sitting_penis_high_noise.safetensors
sitting_penis_high_noise.safetensors
Sex-Insertion_sitting_penis_Wan Video 2.2 I2V-A14B_HIGH.safetensors
sitting_penis_high_noise.safetensors
sitting_penis_high_noise.safetensors
sitting_penis_high_noise.safetensors
sitting_penis_high_noise.safetensors
sitting_penis_high_noise.safetensors
sitting_penis_high_noise.safetensors
Perfect Insertion (back view) HIGH - Wan2.2 - With both hands, she separates her buttocks and sits on the hard penis, inserting it completely inside her pussy.safetensors
Perfect Insertion (back view) HIGH - Wan2.2 - With both hands, she separates her buttocks and sits on the hard penis, inserting it completely inside her pussy.safetensors