For Wan2.2:
Blowjob concept with the man appearing in the frame.
Trained on 20 clips of 3 seconds, 16fps, 512px*512px.
Use this
A woman is performing oral sex on a man's penis.Then customize to your liking.
Workflow is included in the samples.
If you enjoy these models and find them useful, please leave a thumbs up. It really motivates me to keep creating more! Feel free to also share some of the work you generated with them in this page; I’d love to see it.
Enjoy!
For Wan2.1:
Almost all fellatio LoRAs don't show the man, this one does. Go crazy, you animals.
This LoRA generates blowjob/fellatio (sometimes handjobs not sure how) scenes where the man is included within the frame.
Start with the base:
On one side and in the foreground a woman is performing oral sex on a man's penis while he is relaxed and appears on the other side within the frame. That should give you the general concept and the right movement, but then customize to your liking, hair, clothing, general surrounding and so on. I've noticed and I think you did too, that longer prompts work better than shorter ones with Wan2.1, at least this is the case for the videos I am trying to generate.
Works great with character LoRAs.
All dataset were 512*512 and 24fps. 25 videos ranging between 3 to 5 seconds.
Trained for almost 15 hours, 100 epochs and 3600 steps.
Description
FAQ
Comments (18)
nice Lora! Unfortunately the faces of the women are pretty plastic heavy ;P So no cute nextdoor neighbour girls =( Can't find a good negative prompt to get rid of it ... would be awesome if you would retrain it with a modified dataset. The concept itself is aweseome!
I can't say this will help, but I have seen many (including myself) simply use this lora alongside many others: https://civitai.com/models/1385506/detailz-wan-detail-enhancer-for-wan-videos
It tends to play nice with anything, however I have seen it get upset with some I2V lora. Turning it down to 0.5 can be a decent compromise. Trying to beat the "plastic" has been a common theme the last year or so. Only thing I can suggest are tags or phrasing, like "The high-resolution and incredible quality implies the video is reality. detailz. Cinematic. Clear face details. Incredible skin details and natural imperfections."
I really only say this because I personally did not get the plastic effect while using this lora, but have seen it myself.
@makiaevelio543 I think the OP is probably referring to plastic surgery and would prefer training data with more natural looking women (rather than referring to the plastic effect you sometimes get with AI generated images/videos).
Agreed. These women are all plastic surgery-ey and caked in makeup, and the men are all porn star style as well. It's still really good. You can mostly prompt it away, but it does retain the style quite a bit. Would also much prefer amateur style.
Oh I understand now, yes. I was seeing similar faces throughout, but that has been a theme with some trained loras. One has to put in effort in editing the training data to make sure it also doesnt bake the face
Absolute best fucking lora ever. Thanks for making this!
I2V when? Looks pretty cool
works on i2v
Thanks for this!
For anyone wondering: it works for gay bj as well.
You are welcome.
Hunyuan your good loras.
Nice, but the tip of the penis is often really messed up, and as mentioned in another comment, amateur style clips would be much appreciated instead of the stereotypical porn style trained on here.
Amazing stuff man, I just published a Pony LoRA that allows you to create male focused blowjobs, inspired by this.
Thank you.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.