It is trained to produce selfies and pictures that look like real photos. Sometimes it can give silly results, but it can give good results in 1 out of 5 tries. It's up to you to use the word selfie or not. But I recommend it. Try my example prompts. Not needed complicated prompt .
Here is my twitter : https://twitter.com/MetafamilyC follow for the new model
My discord name : MetaGOD#2182
"You can run this model on https://randomseed.co/ with GPU acceleration."
Description
FAQ
Comments (4)
It would be nice to also have a LORA version of this interesting model.
Hmm.. looks very promising. Need to test it ;)
I should say that my prompt consistts body shapes, hair, clothes information, location. I really need it. And thus I think it can kill your selfie "realistic" style. Guess need to do something, but can't be sure that higher weight to word selfie will help, if it really trigger word, of course.
Just an FYI, this model is not available on randomseed anymore



















