FIXPIX: what is happening if the wrong pixels are fixed

Frank Valdes wrote on Sep 10, 1998

Q:  I have having trouble with fixpix. I have made  a mask called bpmask.pl
with the good pixels 0 and the bad one 100. I run fixpix as:

       images = "temp2"         List of images to be fixed
        masks = "bpmask"        List of bad pixel masks
     (linterp = "INDEF")        Mask values for line interpolation
     (cinterp = "INDEF")        Mask values for column interpolation
     (verbose = no)             Verbose output?
      (pixels = no)             List pixels?
        (mode = "ql")           

and none of the bad pixels get fixed. I made the badpix mask at the telescope
with imarith and I converted it to *.pl format by a simple imcopy. If I 
turn on the pixel+ flag in fixpix, the pixels it says it is fixing seem
nothing like the positive pixels in the mask. What is going on?


A:  The behavior you see is because FIXPIX registers masks to data using the
header coordinate system that tracks things like image sections.  However
you created the mask header it has an offset of 522x512 pixels.  This
registering of masks can be very useful but can also less obvious.  Looking
at both your object and mask headers I see that the object header doesn't
have any offset (as given by the LTV keywords).  So you would need to
reset the "physical" coordinate system of the mask:

	cl> wcsreset mask wcs=physical
	
Another way to make sure your mask and image match if you think you made
them to the same pixel system is to change the header of the mask to
be the same as the object using the ARTDATA.MKHEAD routine:

    ar> mkhead mask temp2 append-

---

Last post on Sep 10, 1998