[fitswcs] Quad Cube projections

[Mark Calabretta]

On Thu 1998/07/16 17:15:37 -0400, Doug Mink wrote
in a message to: FITS WCS Mailing List <fitswcs at fits.cv.nrao.edu>

>for that matter) projections.  Based on the changes I had to make in
>the WCSLIB code to get it to work on a test CSC image, I would guess
>that the code in cscfwd(), and probably cscrev(), was never tested
>on a real image.  One problem with these projections is that they
>number the six faces of the cube in a very non-FITS 0-5 instead of
>1-6.  I modified the WCSLIB code to use 1-6 and am wondering how
>people feel that this anomalous numbering system should be treated
>in the FITS WCS standard.

CSC usage was as described by the relevant parties in private communications
to Eric, as noted in G&C.  This includes the cube numbering system, 0-5, whose
main disadvantage, as I see it, is that the faces are laid out in a T rather
than a cross which would give the central face better context.  I translated
the original COBE code into WCSLIB, preserving single precision arithmetic on
the grounds that this is a signature of CSC.  If there are other problems then
please report them to me directly.

Mark Calabretta