Open zonca opened 4 years ago
@nils-erik I see that the WCS keywords are set in the header, however they do not display with DS9:
Can you please crosscheck?
Hi Andrea,
They show up for me.
[image: Screen Shot 2020-02-18 at 2.56.52 PM.png]
On Tue, Feb 18, 2020 at 12:53 PM Andrea Zonca notifications@github.com wrote:
@nils-erik https://github.com/nils-erik I see that the WCS keywords are set in the header, however they do not display with DS9:
[image: image] https://user-images.githubusercontent.com/383090/74777008-8412b780-524d-11ea-929c-10ac17f4f3f0.png
Can you please crosscheck?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/oirlab/iris_pipeline/issues/12?email_source=notifications&email_token=AJPG6OLSITFBWFWB7L3WGITRDRDETA5CNFSM4J3UYEPKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEMEZFJQ#issuecomment-587829926, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJPG6OLZC5LZVTL3644HTPDRDRDETANCNFSM4J3UYEPA .
can you post the image directly to Github? attachments to emails are not supported.
I tested with a HST image and WCS worked fine, so I guess it is something wrong in the file.
it is most probably related to the version of DS9, WCS works fine in DS9 7.5, but does not work on 8.1. Guess it is a problem in the WCS header, I will try to identify the issue.
JWST uses the drizzle algorithm to combine dithered images or mosaics,
this is executed as part of the resample
step which itself is part of the Stage 3 image processing pipeline image3
.
Drizzle parameters are defined in the CRDS, see the schema
@ikashell is it fine if we use the same algorithm and adapt their pipeline to IRIS?
it is most probably related to the version of DS9, WCS works fine in DS9 7.5, but does not work on 8.1. Guess it is a problem in the WCS header, I will try to identify the issue.
I discovered the issue resulting in WCS not displaying in version 8 of ds9. In earlier versions, any invalid algorithm code for the CTYPE keyword in the header defaulted to a linear projection. In version 8 it appears that linear projections require a blank algorithm code (as was always true for spectral axes). I am regenerating the frames in the same directory (it will take a couple hours or so to go through the ~129k GC convolutions) to fix the keyword.
@nils-erik ds9
on my chromebook segfaults with your new files, it works fine (without WCS) on the old files. I will test again on my linux laptop when I get back to the office, which could be weeks, anyway this is not holding me up, the issue is only with ds9
, astropy
works fine with the WCS.
@arunsurya77 do you mind to test one of the files on Galactica in:
/data/group/data/iris/sim/iris_examples/gc_stars/example_field/
can you load it into ds9
? does it display WCS information? what version of ds9
do you have?
Hi Andrea, I tried opening iris_sim_gc_filterKN3_ditherx-20dithery-35.fits iris_sim_gc_filterKN3_ditherx25.dithery20..fits and it opens with ds9 in galactica. the version is 7.5
from @ikashell:
drizzle
is standard algorithm, we should use it as well. Need to test the algorithm and come out with reasonable parameters.
WCS with astropy
works:
In [17]: wcs.WCS(f[0].header)
2020-07-27 18:55:03,591 - stpipe - WARNING - /home/azonca/anaconda3/envs/jwst_dev/lib/python3.6/site-packages/astropy/wcs/wcs.py:466: FITSFixedWarning: RADECSYS= 'FK5 ' /
the RADECSYS keyword is deprecated, use RADESYSa.
colsel=colsel)
2020-07-27 18:55:03,592 - stpipe - WARNING - /home/azonca/anaconda3/envs/jwst_dev/lib/python3.6/site-packages/astropy/wcs/wcs.py:466: FITSFixedWarning: EPOCH = '2019-12-17T00:40:46.48107737302794Z' /
a floating-point value was expected.
colsel=colsel)
Out[17]:
WCS Keywords
Number of WCS axes: 2
CTYPE : 'RA--LINEAR' 'DEC-LINEAR'
CRVAL : 265.197723389 -28.9921894073
CRPIX : 2048.12 2048.12
NAXIS : 4096 4096
@nils-erik I discovered that I need to create a "Generalized WCS" using the gwcs
package,
this package complaints that LINEAR
is not a supported projection, the supported projections are:
'AZP', 'SZP', 'TAN', 'STG', 'SIN', 'ARC', 'ZEA', 'AIR', 'CYP',
'CEA', 'CAR', 'MER', 'SFL', 'PAR', 'MOL', 'AIT', 'COP', 'COE',
'COD', 'COO', 'BON', 'PCO', 'TSC', 'CSC', 'QSC', 'HPX', 'XPH'
Full names of the projections are available at https://docs.astropy.org/en/stable/_modules/astropy/modeling/projections.html
Do you know if LINEAR
is equivalent to any of those?
I fixed this just assuming the projection was TAN instead of LINEAR.
here is the fixes I had to do to the input FITS to be able to use them with iris_pipeline
: https://gist.github.com/568eda8286109688790fe34465dd3e64
Hi Andrea, Sounds good. I wasn't sure what projection was used for the GC coordinates I got from Tuan, but for testing purposes I'm sure TAN is fine. Cheers
On Fri, Sep 11, 2020 at 12:31 PM Andrea Zonca notifications@github.com wrote:
here is the fixes I had to do to the input FITS to be able to use them with iris_pipeline: https://gist.github.com/568eda8286109688790fe34465dd3e64 https://urldefense.com/v3/__https://gist.github.com/568eda8286109688790fe34465dd3e64__;!!Mih3wA!W4uJje4eXUX0GkBgX1Z4TWHVnWhkRQ1eO8t8L_n4KskhyJse7dYMFc2hJIhPcc_0$
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://urldefense.com/v3/__https://github.com/oirlab/iris_pipeline/issues/12*issuecomment-691274916__;Iw!!Mih3wA!W4uJje4eXUX0GkBgX1Z4TWHVnWhkRQ1eO8t8L_n4KskhyJse7dYMFc2hJNYHMaq1$, or unsubscribe https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AJPG6OIV7WMKPFNXAIKNSADSFJ3HNANCNFSM4J3UYEPA__;!!Mih3wA!W4uJje4eXUX0GkBgX1Z4TWHVnWhkRQ1eO8t8L_n4KskhyJse7dYMFc2hJCGBqESe$ .
ok, resample is running to completion on the first image and gives a reprojected output.
@arunsurya77 do you think this looks reasonable?
From Nils: