This seems to have been due to a bad calibration. I eventually got rid of the fisheye and can now see good depth data when viewing my ZED X One stereo pair live in the Depth Viewer.
- It helps to stick the checkerboard onto a nice big white board, with plenty of white space around the checkerboard. I used part of a white melamine timber panel and stuck an A4 9x13 checkerboard to it with no less than 50mm of white space between the edges of the checkerboard and the edge of the timber board. Checkerboard file attached.
- The top left square must be black, not white. If you swap around the target_w and target_h values in main.cpp then you effectively rotate the checkerboard 90deg end end up with a white square in the top left. You will get stuck in the last part of calibration trying to complete the Z-rotation step with the calibration software telling you that you’ve rotated the board too far.
- It seemed to me like the calibration software will happily accept some bad data and proceed to complete a ‘successful’ calibration. But be careful with that, take your time to get good clear pictures of a target on a nice big white surface and plenty of good lighting to illuminate the target clearly.
I am now facing this issue: ZED X One Stereo can't open svo in Depth Viewer - OBJECT DETECTION MODULE IS NOT COMPATIBLE WITH THE ZED