would you consider providing a ensenso_description (or ensenso_support, depending on how many products you intend to support) in combination with the driver?
That would allow you to standardise these frame/link names easily in a urdf, and remove the burden of modelling the camera from your users.
@gavanderhoorn in https://github.com/ensenso/ros_driver/issues/2#issuecomment-344509305: