Ideally, the CaptionRenderer should be able to work with vanilla track objects without having to add any extra bells and whistles to them.