Double facial on the subject's face. Sometimes just a single one ...
Even though it was trained for Wan2.1, it works best using the LoRA for the high noise in Wan2.2 I2V and a cum lora of your likings on the low noise.
I have been experimenting with the following low noise:
F4C3SPL4SH
sh00tz
Facial Cumshot into Hair
Workflows are in the showcase videos.
Description
First LoRA I ever trained. It has flaws in the liquid physics consistency. I have posted this one to see what the community can make of it. Let me know if you want improved iterations.
FAQ
Comments (4)
This was pretty good, but it changed the face quite a bit to the point of not being useable. Can you mask the faces and try again?