walelignmessele

walelignmessele it's using opencv's cv2.findChessboardCorners to detect the chessboard, code here:
https://github.com/luxonis/depthai-experiments/blob/master/gen2-multiple-devices/multi-cam-calibration/camera.py#L135-L138
So it would be best to check opencv's documentation about these functions (here).

Hi walelignmessele
Pre-built devices such as oak-d come with a factory calibration so there is no need to calibrate them on your own, unless you feel like you could achieve a higher accuracy calibration. You do however need to perform a calibration on modular devices (FFCs) since it's you that decides on the position of the cameras and their baseline distance. In short, you only need to perform calibration if you move the stereo cameras around (which is not possible for enclosed devices).

Hope this helps,
Jaka