XUPLX_UglyPeopleLoRA - Small

XUPLX_UglyPeopleLoRA - Small is trained on less than 100 images and 500 steps. It was created as a proof of concept but still holds it's weight against the larger versions.

It may require a higher model_strength if keywords used in your prompt were also used to train ZImage, with the tradeoff being "burning" at nigher values. Changing seeds is preferable, or simply using XUPLX Merge Workflow is preferable to increasing strength.