Parameter support in pinhole camera model

Please provide the following info (check/uncheck the boxes after clicking “+ Create Topic”):
Software Version
DRIVE OS Linux 5.2.0
DRIVE OS Linux 5.2.0 and DriveWorks 3.5
NVIDIA DRIVE™ Software 10.0 (Linux)
NVIDIA DRIVE™ Software 9.0 (Linux)
other DRIVE OS version

Target Operating System

Hardware Platform
NVIDIA DRIVE™ AGX Xavier DevKit (E3550)
NVIDIA DRIVE™ AGX Pegasus DevKit (E3550)

SDK Manager Version

Host Machine Version
native Ubuntu 18.04


We are wondering about the parameter support in the pinhole camera calibration model. As we understand, for distortions the pinhole model only supports the three radial distortion parameters and no tangential parameters. However, there seems to be a bug in which the third radial parameter is ignored

We are wondering if the status of this issue. Is it resolved (in which version?) and if not are there any workarounds?


Dear @tonci.antunovic ,
Do you see the bug discussed in [calibration-graph-cli] Pinhole calibration bug is not fixed in DW 3.5? Could you please check it? If not, please let us know.

Hi @tonci.antunovic ,

DriveWorks 3.5 includes the fix already. Sorry for not informing in the other topic and thanks for checking with us.

Hi @VickNV,

I was going through the DriveWorks 3.5 documentation and I noted that struct dwPinholeCameraConfig still contains distortion polynomial coeff as a 3 X1 sized vector. Could you please let me know whether am missing something here?

Dear @nithin.m1,
pinhole model supports only the three radial distortion parameters. May I know what is your doubt here?

Hi @SivaRamaKrishnaNV,
My doubt here, is it ok to not consider tangential distortion parameters? (Sorry am not familiar with distortions all I want to know is whether it will affect our final accuracy while converting points to world domain)
Also could you please provide me a sample rig file where we are using a pinhole camera? In driveworks examples mostly only ftheta or Ocam models are used.
Thank you,

Dear @nithin.m1,
we’re mainly using the FTheta model to represent our cameras. I guess it depends on the camera whether it is ok to not consider tangential distortion, I think in general we’re working with cameras where this is ok.