Agisoft Metashape

Agisoft Metashape => General => Topic started by: cdc on December 05, 2017, 02:21:18 PM

Title: Camera model details
Post by: cdc on December 05, 2017, 02:21:18 PM

Hi!

I am trying to import the cameras reconstructed with PhotoScan and exported in XML format. Most of its content is clear, only two questions:

• tangential distortion parameters p1 and p2 for the frame camera model seem to be inverted compared to the standard Brown model https://docs.opencv.org/master/d4/d94/tutorial_camera_calibration.html (https://docs.opencv.org/master/d4/d94/tutorial_camera_calibration.html); is that true or is just a typo in the PhotoScan manual?
• the global transform is not described anywhere in the help manual; is it the transform [R*s T] (3x4 in row major format) converting from local to global coordinate system (internal to absolute)?

Thank you!
Dan

PS: great software!!!

Title: Re: Camera model details
Post by: Alexey Pasumansky on December 06, 2017, 06:47:23 PM

Hello Dan,

The order in the manual is correct, so no typo there and it should correspond to P1 and P2 values in the exported XML file (at least, if exported from version 1.4.0). If you are using older version, you can compare P1 and P2 values in GUI with the XML file that you've got on export. At some moment for the backward compatibility the labels of P1 and P2 parameters were swapped in XML.
It means that P1 and P2 order is different from the one used in OpenCV.

As for the global transform, it's from local to global system, so everything seems to be correct here in your assumption.