I fine-tuned an SDXL model on Figures. I probably trained close to 500 models to now be able to achieve this level of quality and generalization on different subjects.