iraf.net

I know this had worked for me in the past when doing MOSAIC reductions but with my current data I cannot get fixpix to interpolate my data well across bad pixels.I have lots of saturated stars in the field, for some reason, some of the saturated pixels have a value of 0 some are above 60000 counts. But it was easy enough to create a bad pixel mask for all of these pixels. Perhaps the screenshots show what I mean (display range 0 to 60000, DS9 color table SLS):Original image:
[img:8194262b07]http://www.aip.de/People/PWeilbacher/fixpix_orig.png[/img:8194262b07]
(After some data reduction, the values near the center of the star are more like -3000 or so but that should not matter.)bad pixel mask (green is 0 and white is 1):
[img:8194262b07]http://www.aip.de/People/PWeilbacher/fixpix_bpm.png[/img:8194262b07]When I try to interpolate using fixpix I get really weird interpolated pixel values:
[img:8194262b07]http://www.aip.de/People/PWeilbacher/fixpix_interpol.png[/img:8194262b07]
The black pixels in the middle are between 15 and 40 counts, while even the darkest blue around is above 6000 counts. I again made sure that the bad pixels in the mask really cover all pixels with low values. So how can that be?[b:8194262b07]Update[/b:8194262b07]: I previously thought that I had converted the small area of the .pl file correctly to a textfile listing all the bad pixels (in lines with x and y as required by fixpix) as a test. But I had done that manually with imexam and the 'x' key and had forgotten two pixels. I now converted the .pl file automatically to such a file and get the exact same weird results.So, I think that I misunderstand something in how fixpix does the interpolation. Does it leave a gap between the bad pixels and the pixels it uses as values for interpolation? If so, how can I switch that off?Is there any method or IRAF task which could achieve the same result?This is with IRAF 2.12.2a on Linux.

Seems like there was something wrong with the fixpix (or something that fixpix depends on) in IRAF 2.12.2.a. I now installed IRAF 2.13-BETA (the redhat binaries) as a test and now interpolation with fixpix seems to work.

Hello,I can help you if you can provide an example for me to reproduce the behavior. If the files are small you can attach them otherwise you can put them at ftp://iraf.noao.edu/pub and notify me. Your description seems clear enough and there is not interpolation buffer. Interpolation is done between the nearest good pixels identified in the mask. While using the text file is reasonable the bad pixel mask is easier and understood by fixpix. One suggestion is that you can try to force interpolation along rows or columns to see what difference that makes. In the mask you do that using mask values of 2 or 3.So let me know if I can help you by looking a case or if you have other questions.Yours,
Frank Valdes
valdes@noao.edu

Frank, thanks for the offer. When I couldn't get it to work beginning of the week it was consistently broken over four different IRAF installs under 3 different user accounts. In the meantime, because it works in the newest IRAF 2.13 beta, I had removed the test files and am not sure which of my science files I created it from. Now, trying it again with an 2.12.2a installation with some random file it works again, too. This is really confusing, but it basically means that I cannot give you test data any more.