This is the SDXL version of https://civarchive.com/models/247837. It was trained on similar dataset, though SDXL requires images of much higher resolutions. It is trained using SDXL base model and will work on fine tuned models, both realistic and anime ones. It worked very well on animagineXL. See sample images for the prompts.

When used on realistic checkpoints, this model has a bias towards Asians, as the training set contains many screen scrapes from Korean / Japanese medical drama scenes. To overcome that, you can add prompts like "Caucasians, blonde hair, blue eyes" or something similar.