Camera Matching adjusts a camera viewpoint so that the perspective in a background image matches the perspective of the digital world (Rotation and location of a camera in a 3D software). This allows you to place 3D objects as if they were part of the background image.
For example ,there exists a software called fSpy, that lets you place vanishing points, representing each of the axis's. fSpy will then generate camera data that you can export to your project, as seen on the right of the video. I'm sure the short video on their website, https://fspy.io/ can demonstrate much better than I can explain in words here, or this image here:
So my question is why hasn't this kind of software been applied to moving video? You can track the points, and then you only have to track 8 points, versus the many you would have to track using standard motion tracking. It also seems to be much more accurate.
Has this been done before, is there a reason why it hasn't been done before, or have I just came up with a groundbreaking idea?