Sure, you can copy the four points you defined in the Planar Tracker to your Corner Positioner:
- Put the Planar Tracker back into Corner Pin mode
- At the bottom of the Inspector for the Planar Tracker, expand "Corner Pin 1" and "Reference Time Positions", and there are your four points:
- They can now be entered into Corner Positioner's equivalent controls:
- It would even be possible to link these parameters, by right-clicking on each set of X&Y values in the Planar Tracker -> Publish, then in the equivalent pair of controls in Planar Transform: right-click -> Connect To -> Planar Tracker 1 -> choose the correct value. That's just FYI; there's no point doing that here as it's quicker just to manually copy the values, and you won't need the Planar Tracker's Corner Pin values again.
All that said, I am slightly confused by the question. Your Planar Tracker Corner Positioner was just eyeballed, yeah? So why would it add any more error to re-eyeball it for the Corner Positioner? That's not adding error onto error, it's doing it again with the same chance of inaccuracy as the first time.
Were you to do this again in future you wouldn't even need to try doing the Corner Positioner in the Planar Tracker. At least not until we know the bug is fixed. You could go straight to using the Planar Transform, and your 4-point positioning would be done once only. You're only doing it twice in this case because you tried corner pinning in the Planar Tracker first, and now it needs to be done again due to this bug.
With regard to your general worry about needing to eyeball the corner pin placement: I'll reply in more detail when you make your separate post, but I'm pretty sure I'm correct in saying that it's entirely normal and correct to have to eyeball the target area/corner pin. That because there is a clear distinction between the search area for the track - as you drew with a spline in the Planar Tracker - and the actual tracking data generated, which you then use to manually place the corner pin.
PS. I've recreated the Planar Tracker bug in a test comp. It appears to happen in any situation where "FG Only" corner pinning is done, and the FG input image is subsequently updated (eg by changing Blur size.) It might be some kind of cache-related race condition, where every time the FG input to the Planar Tracker changes, there's a chance of it outputting a cropped image. I'll raise it in a bug report post later tonight.