Hi,
Is it possible to export camera alignment information including lens distortion values to be used in a 3rd party dense point cloud reconstruction app(like VisualSFM)?
My workflow is:
I have a several cameras in a fixed positions which taking series of shots of different objects. I want to do a camera calibration by RealityCapture for my setup and then export this data through Bundler v3 to be used by VisualSFM for dense point cloud reconstruction so I will not need to do a calibration step each time. Calibrate once - reconstruct many (of course with some limitations)
It seems that RC doing some image resampling (undistortion, principal point translation?, masking? image resizing?) so when I reuse that camera calibration for dense reconstruction with the original images/next image set results are totally messed, means the original images/next image set has to be resampled in the same way to be processed by VisualSFM or any other app? Dense reconstruction with the resampled images(which are generated when I do export registration -> export undistored images true) works fine.
I’ve tried to update calibration.xml by setting undistortImages=“0” undistortPrincipal=“0” but no luck.
Is there any other options I should set in calibration.xml to export camera alignment which can be used for dense point reconstruction with the original (non-resampled) images?
Thanks in advance.