Crotchface, pussyface, facegina, whatever you call this.
Can do closeups, full-body, nude, clothed, face covered or uncovered.
Use trigger word "crotchface". Lora strength 1 recommended.
Trained on a batch of 40 or so photo edits. Intended for use with photorealistic models. Other styles may work but are not tested.
Update: Wan2.2 T2V A14B
You need to use both the high noise and low noise LoRas together. See the embedded workflows for examples.
Update: HunyuanVideo Version
The Hunyuan Video Lora was trained on largely the same dataset as the Pony XL Lora (52 images, 2 videos), using default parameters from diffusion-pipe.
For best results, load the Lora with the single blocks at half weight, and the double blocks at 1.0 weight or more. Some videos in the gallery were generated with additional upscaling workflows and don't include the original workflow.
Description
FAQ
Comments (18)
what in the name of all that's holy
haha my reaction exactly!!
When neck-vagina?
The king has returned!
Y’all are worrying me
So, can she give a bj?
What in the actual fuck?
"we can't I'm on my period"
Just gimme some head then.
This is the most creativity Lora I've seen. I'm curious how trained it. nice job!
Please make more videos like this. I love crotchface womens❤️🔥💋.
Amazing work! Were there any tutorials or guides you used when making the Wan loras? You've inspired me to try it out myself.
Ok, that's enough scrolling, gnight folks.
How did you even train this? What kind of data did you use?
The training data was originally all photo edits that I had done (about 40 images). I used that to train the pony and hunyuan loras, then used those to generate a few more high quality images and videos for the Wan2.2 lora.
This is what high quality lora equates too, and this is not sarcasm.
Is there a possibility that you could make an illustrious version of this model?
Maybe? I tried that once with the CivitAI on-site trainer and the results were not good. I'll think about giving it another go. The dataset is all photos, not sure if that's problem for Illustrious.
@harrisonb48323 That should not be a problem -- illustrious is just a better version of Pony