SDXL has a higher base resolution, I wonder if it's better to have more zoomed out dataset with more details or shows the entire characters body instead of close ups.

So far, all of SDXL checkpoints and LORA's seems to have washed out details, specially with close ups.