[daip] Channel 0's

[Eric Greisen]

Lynn D. Matthews writes:

 > I was wondering how the new FILLM selects the channels it uses for the
 > computation of the new Channel 0. If the answer is "inner 75%", then my

It is:       FR1 = (NFR+1) / 8 + 1
             FR2 = NFR - ((NFR+1)/8)
where        NFR = 15, 31, 63, 127 etc

Note that you can set a bit to get the on-line channel 0, but that is
computed when the data have more bits than they do on tape and before
they are Hanning smoothed

 > second question is how this differs from the default "inner 75%" used by
 > AVPSC. I was curious, since this latter program (used in default mode) has

Oops - I fogot to fix AVSPC's default, currently:
            CHNSEL(1,1,K) = (NINCH+1)/8
            CHNSEL(2,1,K) = ((NINCH+1)*7/8) - 1
            CHNSEL(3,1,K) = 1

 > the property of reassigning the reference pixel to a non-integer value,
 > whereas FILLM does not, and a comparison I've done between the two
 > resulting CH 0's shows they are not identical.

       They should be subtly different until I fix the default in
AVSPC (today).  As for reference pixel - I forgot to check what FILLM
does and so it does what it did before (I think) and before it had the
pixel values of the not-yet-fixed AVSPC.  AVSPC does try to do a
careful job of the channel increment and reference pixel

Eric