[iaufwg] FITS conventions

[Mark Calabretta]

On Tue 2006/02/14 08:37:06 BST, Francois Ochsenbein wrote
in a message to: Mark Calabretta <mcalabre at atnf.csiro.au>
and copied to: iaufwg at nrao.edu

>Thanks for the "Colour FITS" paper -- quite interesting despite I coulnd't
>get the colouts on Fig.1, and Eqs. 8-10 seem to be missing.

Bonjour Francois,

What did you use to display the it?  Can you see the colours using 'gv'?

The PostScript is coded in accordance with the procedures described in
Sect. 4.8 of the "PostScript Language Reference Manual" 2nd ed. using a
CIE (1931) XYZ colour space by default, although it can easily be
switched between that and sRGB or CIE L*a*b* by commenting/uncommenting
a couple of lines.  I've appended these variants; they should look
identical (but not necessarily true) on whatever device you use to
display them.

It's not just Eqs. 8-10 (trivial) that are missing, great chunks of
text are as well.

>It is certainly important to define accurately the colours in FITS,
>in a device-independent manner. My fear is that this could require another
>decade until the final agreement... for a recommendation about the contents of
>the CTYPEx keywords. BTW I was not aware of the 'sRGB', 'CIELUV' or 'CIELAB' 
>conventions, they would most likely be worth being publised in the FITS
>convention pages  (http://fits.gsfc.nasa.gov/fits_conventions.html)

The main conclusion, at least for now, is that colour should not be
represented as a byte-triple (or quadruple for CMYK) using BITPIX
because setting BITPIX=24 only allows one colour space to be supported.
CTYPEia allows much greater flexibility, both in supporting any number
of colour spaces and in allowing different storage options for the
colour components, e.g. as byte-triples on the first image axis, or as
separate colour image planes on the last axis.

>About using other packages -- it is serious to propose using tiff or
>GeoTIFF ? Should we seriously think about moving away from FITS because
>other formats have reached a similar maturity as FITS ? A fundamental
>question for the future !

My answers to those questions are given in the paper itself.

Regards, Mark