This is my first lora.
Inspired by
https://civarchive.com/models/1518243/wan-22-wan-21-taker-pov-fellatio-i2v
This creates a video that implies viewer being in woman position. FPOV - female point of view.
Wan2.2 T2V 14B High/Low; civtai dont let me upload files seperately; "same size"
I used k3nk awesome workflow to create the videos.
Anyways, zip file contains high,low noise.
Description
Final
FAQ
Comments (20)
Difficult to change the Male to a futa, its very specific, but nicely done , amazing that this works at all.
Hey 🙌 could you try training the opposite concept? A camera on the penis shaft about to penetrate.. whatever (mouth, pussy, etc...) Think it's possible? Thanks
There was a snorri cam Lora for 2.1. might be worth a try?
@rustybutler Snorri cam? I'll look for that, no idea what it is
Amazing lora, make armpits licking please wan 2.2
dislike if you gay
it won't get funny. no matter how many times you spam it.
Can I dislike if I'm het?
@MMOFan he had a different text when I disliked it. the kid loves to edit his spam.
@janssenbutton gay kids are funny, by definition, I think
This one will be popular with a certain group. For others tho, maybe an idea for a new lora could be:
A close up of HER entire face as it goes in from the MAN's perspective?
yeah im surprised this is still missing
It's my own fault that I pay for the Internet.
I figured you were trying to upload the same file for both high and low like some people tend to do, but they are in fact distinct files, just exactly the same filesize
Yup , I had to double check that. High/Low noise was trained at the same time on Runpod. Guess that is how that happens; since they get equal time training.
Well, it definitely works. Maybe a bit too well in the sense that it is very rigid and seems to take prompts as gentle suggestions and do its own thing most of the time. I was hoping for more flexibility but you had one goal and you nailed that very nicely. Great stuff.
Thank you , It something I have felt too. Its not very flexible - also due to its training data; you always/mostly get some hair. which sucks.
I has been a learning expierance.
Can I ask how many videos you trained on? Size? Workflow?
I tried making something similar and it messed up nearly every time.
45 video clips + captions, 5 second clips ( 512x512 pixels ) cutout from larger videos with capcut, I use the K3NK workflow as in comfyui "workflow" / if you mean "workflow" as in training I trained on runpod hearmemanai wan guide.
@randyrocketcorp497 ahhh amazing. Thank you. I was training on too few clips. 45! I wasn't using enough clearly.