iraf.net

Hi,
I have been trying to fix several features on a flat frame, which move due to flexure in the telescope and hence cause spurious positive features when they are applied to an image, with the task fixpix. I have an badpixel mask that I generated with a custom algorithm that completely covers each of the spots that I want to remove. The problem is, when I apply the mask with fixpix, the spots become more like L's. When I tried changing the parameters to force column interpolation and they became vertical lines, and when I forced line interpolation they became horizontal lines. It looks very much like fixpix somehow applied the bad pixel mask to the wrong locations. This sounded like the fixpix error that was fixed in version 2.1.3, so I deleted IRAF from my system (an intel macintosh) and upgraded to version 2.1.3 beta2, which I obtained, along with an installer, from macsingularity.org, but fixpix still doesn't work right. I would love to show you images if that will help to clarify the problem, but I think I will need to send them over e-mail. Any help you can offer would be great.~Sean

Sean,There may well be a bug, but there are no known issues with FIXPIX so it could equally well be a bad pixel mask. It would help to be able to see the data and the parameters you were using. You can upload data to ftp://iraf.noao.edu/pub and post a notice here along with the parameters you're using (e.g. "dpar fixpix"). Be sure to include the pixel mask and we'll have a look.Cheers,
-Mike

I'm having a little trouble getting into your ftp server (probably just because I'm rather novice at UNIX). When I enter the command ftp ftp://iraf.noao.edu/pub, I get the output:Connected to tucana.tuc.noao.edu
220 Welcome to NOAO-Tucson IRAF/NVO FTP service.
331 Please specify the password.
230 Login successful. Have fun.
Remote system type is UNIX.
Using binary mode to transfer files.
200 Switching to Binary mode.
local: pub remote: pub
500 Unknown command.
227 Entering Passive Mode (140,252,1,86,220,97)
550 Failed to open file.
221 Goodbye.And during all of this I didn't enter a single command. What am I doing wrong?

The 'ftp' command on OSX also acts as a download client. Try using just[code:1:4675cebcea]> ftp iraf.noao.edu
....login as 'anonymous', give email as passwd
ftp> cd pub
ftp> binary
ftp> put myfile.fits
ftp> put myfile.pl
ftp> quit[/code:1:4675cebcea]-Mike

Ok, I've sent four files. badpixelmask.fits is my bad pixel mask in .fits format. Due to the way it is generated, it cannot reside as a .pl file (it is usually a .imh/.pix file). Maybe this is part of the problem?fixpixparameters is the textfile output of dpar fixpix.originalflat.fits is what the image that I masked originally looked like.maskedflat.fits is what the image looks like after it had the bad pixel mask applied to it.Thanks!
~Sean

Sean,I can't fully explain your data, but I don't think there's a problem in the task. I noticed a few things: - The 'L' shapes in your maskedflat.fits file appears to be the result of a shift between the mask and the input file and not a case of bad interpolation. There's no header information in the files so I can't say how that might have happened.- Your badpixelmask.fits file appears to be mostly values of 1 or 200 -- the task can take a standard image as a mask file but assumes "good" pixels are flagged with a zero value- By subtracting 1 from the mask file I was able to run FIXPIX with that new mask either as a FITS image, an IMH, or a pixel-list .pl file with no 'L' shape in the final file. In my result there is also a 1 pixel on all but the left side of the image, in your maskedflat.fits file the "border" is several pixels wide on the left side.Are you sure there's not something funny going on in the mask creation task?Cheers,
-Mike

Hi Mike,
It is certainly possible that something odd happened during the mask creation. I had also noticed that the 'L' appears to be the result of a shift between the mask and the flat, but that does not make much sense since I used geomap and geotran to align the two, and then checked the alignment visually by blinking the two images. I am surprised, however, that you found that there were no zero values in the mask, I thought I had explicitly set them to zero, so perhaps something that I am doing between when I set their values and when I apply the mask is changing them to one. Could it be that geotran is doing an interpolation that causes this? Anyhow, if you can get it to work on your end, then the problem must be on mine and not a problem with FIXPIX.
One more thing, if you wouldn't mind. Since I've upgraded to 2.1.3beta2, rfits seems to have stopped working and gives me the "Abnormal termination of child process" error. I know that you have dealt with this before with Marek42 and posted a fixed binary, but the link appears to be broken. Is there some other place I can download the fix?Thank you very much for your time and quick responses!
~Sean

The link is correct but somehow the period in the sentence got to be part of the URL. The link is: https://iraf.net/ftp/pub/x_dataio.e.MACINTELGEOTRAN could be causing the problem; The mask file you gave did indeed have zero values, but only a few. The majority of what I took to be good pixels were set to one. Note that simply subtracting one from the mask gave me good results, I didn't need to shift anything.-Mike