[fitsbits] Re: FITS vs. TIFF (or other image formats)

[Rob Seaman]

Paul Schlyter writes:

> FITS doesn't store images in a particularly efficient way though,
> mostly because it's an ASCII format.  However being an ASCII format
> means it's easily transported between widely different computers.
> And I believe that was one major aim of FITS: to be portable between
> as many computers as possible, including several mainframes which
> nowadays are considered obsolete.

Only the headers are ASCII, the data records are typically binary values
that are packed as efficiently as any other storage format.  There is
currently no FITS standard for data compression - but also no reason
compression can't be applied externally to entire FITS objects.  Not
ideal - but not a bad workaround.  Note that astronomical data are
often rather incompressible in any event due to the high density of
information content.

The portability of FITS is through time, not just between current epoch
computers.  It is a feature to be prized that FITS continues to run on
obsolete legacy equipment.  Today's P4 Linux boxes are going to be
obsolete soon enough.  Assuming our great^N grandchildren can find a
drive able to read our tapes or disks, decoding the contents of the
media should be straightforward.  This will likely not be true of
commercial data formats that may be defined by specific and changing
software version numbers - even assuming the software vendors remain
in business decades hence.

Astronomy is a field in which current practitioners find the need to
consult data "published" hundreds or thousands of years ago.  We need
an industrial strength data format like FITS.  Like other industrial
products, FITS may be less slick than short lifespan consumer products.
It is, however, far less disposable.

Rob Seaman
National Optical Astronomy Observatory