We can extract a general conversion factor from detected photons to GAIA magnitudes, however, the variation in photometric quality is large enough to obscure any relation based on zenith angle; for that, we would need photometric conditions. 

I've processed the AG data we have which is both focussed, and had a good enough transformation to produce a match with the Gaia catalogue. For each frame, I fit a linear function with outlier rejection to the brighter sources between log10(flux) and Gaia g band magnitude, where the flux is the instrumental flux returned by sextractor.  

We don't see any systematic variation in the conversion based on the camera ID, the side of the detector (glass vs no glass), or the altitude of the observations.  The variation in photometric quality overwhelms any of these factors. 

The first plot shows an example of a single fit; the colour coding is by altitude, and the line shows the fit/fitted region. 

The second plot shows all the individual fits (one fit per exposure ID) on a single plot. From this plot, I selected the median slope as an approximate fit to the Gaia magnitudes. 

Code is updated and pushed to branch for testing. The coefficients of the fit are stored in the agcc.yaml file; the updated version is in the corresponding branch of pfs_instdata. Database write has not been added yet, as this will require a schema update. 

The conversion formula was implemented for the 2023 April/May run. Analysis of the new data will be done under the new ticket (INSTRM-1979).