Hello!
I’m trying to calibrate mapping on cyclorama shape with 6 projectors, since they overlap each other i need good aligment + blending map. I tried to use CamSchnappr, but since i use short trow projectors (Optoma GT1080) with non ajustable lens shift i can’t get accurate alignment for my setup, even with 12+ points in CamSchnappr. 3d model that i use is quite accurate and simple.
It seems, that problem is wrong principal point’s guess (sure, i turned off fix principal points in advanced tab).
Since i know basic projectors characteristics (FOV, lens shift - or fx, fy, cx, cy in OpenCV ) i would to define them manually, as i understood, OpenCV camera calibrate allow’s me to do that with CV_CALIB_USE_INTRINSIC_GUESS flag which i guess this is Intrinsic guess parameter of CamSchnappr’s advance tab.
But when i turn it off i get error inside CamSchnappr network: camSchnappr/camSchnappr/secondScreen/script1 has error:
(-5) For non-planar calibration rigs the initial intrinsic matrix must be specified in function cvCalibrateCamera2Internal
Could you please explain how can i use CamSchnappr if intrinsic parameters are known, where should i put that parameters (fx, fy, cx, cy and other) and why i get that error message no matter what setup i have?
Thank you!