The biggest tell for me after ears, hair and backgrounds was hard-lighting. None of these generation models 'learn' shadows with point-like lighting.