Hello doppelganger,
There are two general matching-alignment modes in PhotoScan:
- image matching is performed by PhotoScan and the camera locations is estimated based on this information (by default the calibration parameters are also calculated by PhotoScan, but it an be disabled), The procedure is commonly run via Align Photos dialog,
- camera locations, orientation information and calibration data is imported to PhotoScan (for example, via Import Cameras option or using Python) and PhotoScan is calculating only matching points according to the provided information (Build Points operation), so neither exterior, nor interior orientation information is adjusted following this approach.
In principle, when the first approach is followed, you can load the reference information to the Reference pane (like you were doing it originally), specify the high accuracy values for camera location/orientation fields and optimize the alignment.
But I do not recommend to use "zero" accuracy values, as it could lead to the invalid results.