Doggystyle with a female face.
It not good at vertical composition. (It is trained with square pictures)
Start to try with
<lora:pnx_doggyFaceFocus:0.7>
CFG scale = 5
Description
FAQ
Comments (9)
trained only with asian models?
just add caucasian in positive and asian in negative
No, it is not only Asian, but mainly with Asian.
"race" is not the issue of Rola. We can express the images from the model by prompts like "cuasian".
"blonde_hair", "blue_eyes", "dark_skin" are the other examples.
I posted new images with dark skin and blonde hair girls, Check the prompts.
But, I found a problem.
Dark skin girls have white hip!
My Rola has strong bias at the color of hips.
I'm going to fix it at next update.
Seems like it was only trained on Asian girls, making it unsuitable for use with other character LoRAs.
What tool did you create it with?
Lora easy training scripts and Google!
How did you make the male's face visible?
Not good for diversity, it will heavily weight on Asian faces
Don't work, the results are horrific messes of half formed bodies, even when using the prompts from the sample images.