My first attempt at training Z-Image on a 3060. It feels almost as easy as training SD 1.5 used to be! Will definitely try to make more LoRAs -- open to requests.
It can have trouble with more than three girls -- you might have to try generating a few times. You can get better and more consistent results by being very deliberate and precise with your prompt for each girl.
Upcoming V1.3:
Differentiating between sucking, swallowing, and "throating" a penis.
Changed "making eye contact" prompt (which Z-Image has trouble with) to "staring at camera".
V1.2 Changelog:
Added some images of just a POV view penis to training set
Went back to older, harsher training settings -- but using only a partially trained output for better flexibility.
V1.1 Changelog:
Fixed typos in some captions (yes, it's all manually captioned)
Added more asians and latinas into the dataset
Set rank to 16 from 32
Set learning rate to 0.0003 from 0.0004
Description
Fixed typos in some captions (yes, it's all manually captioned)
Added more asians and latinas into the dataset
Set rank to 16 from 32
Set learning rate to 0.0003 from 0.0004
FAQ
Comments (1)
This model has a very strong influence but is easily balanced out by adding even an unrelated LoRA, to push some of its flavor aside in order to expose what other things might be hiding inside.
So, fiddling with different this'siz and that'siz, the POV angle (that the model was apparently trained with) seemed to step aside and allow for a couple of disparate offerings, one of which seems to work well enough so I posted an image of the result.
The angle presented was consistent, it pretty much never failed, but I had to cherry pick quite a bit (one LoRA I chose was goofing up the shoes and the fact that the Z Turbo is distilled probably doesn't help us much).
Thank you for this, POV is actually one of my favorites but I accidentally bumped into something else interesting so thought I'd share.
I used version 1.1.