Not to mention that by telling it the wrong height it would mess up the laser focusing. Although I guess you could fudge the height, lay out everything, then set the height back before clicking “Print”. But if the only thing you’re fudging for is the lid-camera to crumb-tray then I could imagine this could conceivably work (I’m curious enough that I’ll test it tomorrow). One way to see would be whether straight lines in the stock become straight lines in the preview image.
The calibration error is not a linear transformation (because it’s a fisheye lens and camera distortion). But it can be still described mathematically and in theory you could extract the parameters for an approximation of the error function from a print like what I showed above. This isn’t enough to fully calibrate the lens from scratch, but it would be enough to improve it.
I’ve seen references on the forum to people who’ve had their calibration fixed “magically” by Glowforge by what sounds like them analysing the before/after bed images from the gift-of-good-measure print. Would love to know from Glowforge staff if this is true?