We have been using a ChArUco board (4 ft x 4 ft) and letting Metashape do photo alignment on image sets we gather on the board. (
https://docs.opencv.org/3.1.0/df/d4a/tutorial_charuco_detection.html). When gathering photos, we circle the board and gather 40 images from various angles with the camera as normal, and then 32 more with the camera inverted (e.g. upside down). Camera settings are fixed with the focal length at infinity (max), shutter at 1/80, and ISO set to where EV is 0. When aligning photos, we use the highest accuracy and uncheck adaptive camera model fitting. After images are aligned, we can see the calibration results in Metashape and save the .xml file for use later.
With Metashape, there appears to be a change in that a user can now use a precalibrated camera (.xml file) and fix only certain parameters. If adaptive camera fitting is turned on during photo alignment, does it only estimate the parameters that are not "fixed" in the camera calibration settings?