I would love that to work like this, because I can then use camera track / projection as a stabilizing method for paint/cleanup etc.
That is exactly what I try to do
I have a camera track with proxy geo.
Do the camera project like I said earlier.
Use the catcher tool to feed to the geo(with UV)
Use merge 3d to merge camera and the geo
Then use renderer3d, set up the render image to square(ex 1024x1024)
Done