There's already quite a few similar models on here, but they tend to suffer from the old "same face" issue: That is, they're overtrained on a limited number of faces and these shine through when prompting unrelated people.
My goal with this LoRA was to to generate facials that don't suffer from the same face issue. Hence it's been trained on a vast variety of subjects: Young (18+), old, big, small. Every possible race, creed and ethnicity was included. In > 150 high-res training images, not a single person was featured more than once.
If it's got a face, chances are this model has seen someone's nobslurry on it. This is without a doubt an equal-opportunity splooge-fest.
Was I successful? Not entirely. But the model does add a bit of variety.
It comes with a few extra control parameters:
closed eyesunhappyfor those cum haters