Real sense cameras in our lab D435/D415 are factory calibration for intrinsics (and extrinsics between RGB and D camera). However, the intrinsics set the distortion parameters to 0.0, i.e a distortion model is not considered as mentioned here
Assuming there is zero distortion might not be the best approach for robot-camera calibration and mulit-camera calibration(at least in Kalibr, intrinsics are calibrated and then extrinsics).
To deal with this, we could first calibrate the intrinsics, and then use the distortion parameters in the calibration code here and here.
Unfortunately, we can't think of a way to update the intrinsics in the camera firmware yet. Doing this makes it easier, to directly read from ../camera_info topic if using ros driver for the camera, or read from pyrealsense SDK like reading the camera matrix here. Therefore, right now, we are advising to manually set the distortion parameters estimated using Kalibr.
Real sense cameras in our lab D435/D415 are factory calibration for intrinsics (and extrinsics between RGB and D camera). However, the intrinsics set the distortion parameters to 0.0, i.e a distortion model is not considered as mentioned here
Assuming there is zero distortion might not be the best approach for robot-camera calibration and mulit-camera calibration(at least in Kalibr, intrinsics are calibrated and then extrinsics).
To deal with this, we could first calibrate the intrinsics, and then use the distortion parameters in the calibration code here and here.
Unfortunately, we can't think of a way to update the intrinsics in the camera firmware yet. Doing this makes it easier, to directly read from ../camera_info topic if using ros driver for the camera, or read from pyrealsense SDK like reading the camera matrix here. Therefore, right now, we are advising to manually set the distortion parameters estimated using Kalibr.