[fitsbits] Any bad pixel mask convention?

[Lucio Chiappetti]

On Wed, 4 Dec 2013, Richard J. Mathar wrote:

> Is there something like a bad pixel mask FITS standard that stores bad 
> pixels with a boolen value or perhaps a float where 1=good, NaN=bad in 
> FITS images?

Definitely there is nothing in the standard (not its business).

I am not aware of any *convention* which is registered or of general use, 
I suppose each project uses its own.

If I were to store something permanent or semi-permanent, I won't use 
something contrived and space wasteful as float 1/NaN, but just the 
simplest approximation to a boolean map, an integer BITPIX=16 1/0.

(quite a long time ago I had Fortran routines which emulated a bit array 
(matrix) but there were never stored to files)

If the map is of transient nature, I won't worry about the size, but just 
use a 1/0 or 1.0/0.0 or 1.0D0/0.0D0 map analogous to the other images I'll 
be operating on (I generated some masks with some ftools operating on 
other images ... for instance the simple way to initialize a mask at the 
wished size is to subtract an image from itself)

-- 
------------------------------------------------------------------------
Lucio Chiappetti - INAF/IASF - via Bassini 15 - I-20133 Milano (Italy)
For more info : http://www.iasf-milano.inaf.it/~lucio/personal.html
------------------------------------------------------------------------
Wir muessen wissen / wir werden wissen
          Inscription on David Hilbert's grave  http://tinyurl.com/3e5hfl
------------------------------------------------------------------------