jmiguel |
09/17/2014 10:49AM (Read 2886 times)
|
|
|
Status: offline
Registered: 12/12/2007
Posts: 14
|
Hi everybody,
I am running psfmeasure with a 32x32px image with a unique star in the center of the field, and I get the next message:
Warning: Pixel subscript out of bounds.
and thus no other result is obtained.
Do you have any suggestion to avoid this warning/error message from psfmeasure ?
Thanks,
Jose
|
|
|
|
fitz |
09/17/2014 04:02PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
If you're using DS9 v7.2 as the image display then the problem is a known bug in the cursor readout in that version, the only workaround is to use a different version of DS9 or XImtool as the display server. In other recent DS9 releases you must use
setenv IMTDEV inet:5137
before you login to the CL to force the use of inet sockets for cursor reads to work properly.
|
|
|
|
jmiguel |
09/17/2014 04:36PM
|
|
|
Status: offline
Registered: 12/12/2007
Posts: 14
|
Yes, I had read something about a similar problem on imexam, so I was using DS9 v7.3b5. Now Imexam works with 32x32, but psfmeasure does not.
I have just tried your suggestion ( setenv IMTDEV inet:5137 ), but the problem continues.
|
|
|
|
jmiguel |
09/17/2014 04:41PM
|
|
|
Status: offline
Registered: 12/12/2007
Posts: 14
|
Quote by: jmiguelYes, I had read something about a similar problem on imexam, so I was using DS9 v7.3b5. Now Imexam works with 32x32, but psfmeasure does not.
I have just tried your suggestion ( setenv IMTDEV inet:5137 ), but the problem continues.
You can find the sample image here: ftp://ftp.iaa.es/pub/jmiguel/osn/test32x32.fits
|
|
|
|
fitz |
09/17/2014 06:32PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
The same cursor problem appears to be in the v7.3 and v7.1 DS9 versions, I had to go back to a 7.0.1 release before it would work. You can get older versions of DS9 from their archive at http://ds9.si.edu/archive
|
|
|
|
jmiguel |
09/18/2014 06:41AM
|
|
|
Status: offline
Registered: 12/12/2007
Posts: 14
|
Quote by: fitz
The same cursor problem appears to be in the v7.3 and v7.1 DS9 versions, I had to go back to a 7.0.1 release before it would work. You can get older versions of DS9 from their archive at http://ds9.si.edu/archive
I still have the same problem with 7.0.1; honestly I think the problem is in psfmeasure with that kind of 32x32 image having a star with a FWHM~7px in the center of the field. Please, could you test the image ftp://ftp.iaa.es/pub/jmiguel/osn/test32x32.fits with your setup ?
Thanks,
Jose
|
|
|
|
fitz |
09/18/2014 05:32PM
|
|
|
Status: offline
Registered: 09/30/2005
Posts: 4040
|
In this case I agree PSFMEASURE is doing something weird: Using a cursor file (i.e. no DS9 at all, just a text file of coords) I can reproduce the message (at times). Part of the problem is that the boundary size used in the task exceeds the image size and so some requested coords are indeed out of bounds. I'll ask Frank to have a look ....
|
|
|
|
jmiguel |
09/19/2014 03:53PM
|
|
|
Status: offline
Registered: 12/12/2007
Posts: 14
|
Yes, I have also checked that the problem also happens (always says: 'No input data') with a cursor file in a non-interactive script. So, It looks like it does not depend on the DS9.
By the way and concerning the cursor file and the script:
Do you know if there any way to avoid "the black window" that arise each time a cursor file script is executed ?
|
|
|
|
valdes |
09/19/2014 05:56PM
|
|
|
Status: offline
Registered: 11/11/2005
Posts: 728
|
Hola Jose,
We believe this is a pathological case. The PSF width is large and the image size is small. What the program does is reflect the boundaries of the image to avoid biases for stars near the edge. But in this case we think the reflection is producing a signal from the wings causing the centroiding to go loco. The code fix, for a future version, is to use a different type of boundary extension. For a workaround you can embed the image into a bigger canvas or use blkrep or magnify to make the image bigger so that the 100 pixel boundary reflection no longer includes enough signal to confuse things. To embed a smaller image in a larger image you use imcopy with an image section and you can make a dummy image of a desired size in various ways including artdata.mkpattern. Let me know if you need more details.
What is still unclear is that I did not have a problem with your image on a 32-bit Linux system but did have it on a 64-bit Mac. I don't know what system you have but there can still be some bug with 64-bit systems in the boundary extension. So let us know on what OS you discovered this problem.
Yours,
Frank Valdes
|
|
|
|
jmiguel |
09/19/2014 08:39PM
|
|
|
Status: offline
Registered: 12/12/2007
Posts: 14
|
Hola Frank Valdes,
Thanks for your time and your explanation about what is happening. I will try your suggestions and let you know my results.
Concerning the running system, I discovered the problem on a 64-bit openSUSE 13.1 system, running IRAF Revision 2.16.1, however it also happens on 64-bit openSUSE 12.2. Please, let me know if you need I run some test or check on my computer.
Un saludo,
Jose
|
|
|
|