This is the SDXL version of my pack of female faces (intended to fix the same-face phenomenon.) Since I'm not that big of an SDXL user, this will not be updated as frequently as my Virtual Goddess SD1.5 Female Face Pack though I may if there's more interest.
Please see the "About this version" section to the right for the the respective LoRA version to get the trigger words for each of the faces in the LoRA. Note that the latest version was trained on 200 different faces and this has 200 different trigger words, one for each face.
Description
Note: this LoRA contains 25 trigger words, each for a different face:
ZMXLelena
ZMXLcaroline
ZMXLdina
ZMXLlaurie
ZMXLmichelle
ZMXLkonstandina
ZMXLangeliki
ZMXLcatherine
ZMXLdelphine
ZMXLvilde
ZMXLjessica
ZMXLelif
ZMXLchelsea
ZMXLelaine
ZMXLhelle
ZMXLpetra
ZMXLwenche
ZMXLmary
ZMXLnancy
ZMXLangel
ZMXLashley
ZMXLkirsten
ZMXLgalina
ZMXLamy
ZMXLleslie
FAQ
Comments (3)
I think your dataset caused most of them to look a bit ugly. Getting ugly is fine, if prompted for it, but I would imagine this lora judging by the title should at least go the range into beautiful, and not give unintentionally ugly and very asymmetrical faces.
Well, maybe I should stick with 1.5 where everything's turning out beautiful without much effort.
I disagree with the other comment. With 25 face choices and hundreds of models to choose from, you're bound to eventually find a combination that produces beautiful faces with this LoRA. Of course, some of the 'ugly' results aren't necessarily bad either.