By not showing anything do you mean it is displaying text that is like "scan image" but it doesn't show you what the image looks like? If so this is how it currently works on all devices except Hololens. but if you know what the image is and you get it in the frame View will track it very quickly in my experience, much more robust than model tracking.
