The current pfsConfig (soon to be pfiDesign) specification contains an array of five bandpass magnitudes, to be used by the QA software during observing. The filter names are not actually declared but default to the obvious ones from HSC.

We need to specify this better.

• We can filter the 1-d spectra to match most bandpasses, but some SED needs to be provided for any we use. Some photometric systems will be precisely defined (HSC), and mean wavelength + width is probably good enough in most cases.
• We may need to apply seeing corrections during observing (stronger effect towards the blue). Again, some photometry sources will have better information than others.

With those in mind, we are proposing the following for pfiDesign, or at least for discussion:

• pfiDesign objects can list magnitudes in any bandpasses which are known to the QA system: that is where the values will be used. The DRP/QA does/will have a flexible filter class.
• We require AB magnitudes.
• we still need to figure out how to specify fluxes under different seeing conditions. Or decide that it does not matter.

Perhaps we should have an extensible list of known and supported filters? Then either require that pfiDesign filters be in that list, or allow an additional HDU to specify the filter properties? The latter seems messy, but would be self-contained.