niriss image3 crashes with "MalformedPolygonError: intersection: find all intersections" #8513
Comments
|
Comment by Howard Bushouse on JIRA: Very bad looking step parameter values being assigned by the active pars-tweakregstep ref file for NIRISS images. Wouldn't be surprised if that's what's causing the failures. |
|
Comment by Howard Bushouse on JIRA: Hien Tran Could you make the image3 asn file and the cal files listed therein available somewhere for SCSB to do testing? We want to make sure to use the exact same inputs as used in Ops. |
|
Comment by Rachel Plesha on JIRA: Kevin Volk, Jo Taylor, André Martel, Tony Sohn, Paul Goudfrooij , I wanted to bring this ticket to your attention. There were some thoughts at the JP meeting yesterday that this might be due to some parameters that are set to bad values, specifically in the search radius and tolerance values when using tweakreg with the gaia3 absolute alignment in the parameter reference file. I'm not sure if any of this will be affected or fixed by the upcoming changes using the IRAF starfinder, but please chime in if you have any ideas about this error or upcoming updates to the values. |
|
Comment by Kevin Volk on JIRA: The issue here seems to me to be that the image size is too small and there are not enough stars to make the tweakreg catalogue. This example is a photometric calibration observation with image size 128x128 pixels. The image attached shows the field in the rate image. There is just one star, the photometric standard, present. Apparently tweakreg is not finding this star (which is a problem in itself given the image...) and produces no catalogues for matching and so things crash. The question is why it is crashing currently as opposed to in previous runs. The revisions to the tweakreg parameter files should not cause problems in an image like this. Nonetheless that may be the issue. Howard commented that the tweakreg parameters seemed bad; but which ones? |
|
Howard Bushouse i put the data in ████████████████████████████ |
|
Comment by Mihai Cara on JIRA: Are there specific instructions that could help me reproduce this issue? I run: pip install -r reqs_macos-stable-deps.txt```
where reqs_macos-stable-deps.txt was downloaded from here: <https://ssb.stsci.edu/releases/jwstdp/1.13.3/reqs_macos-stable-deps.txt>
I then run the step using strun, creating TweakRegStep object or creating Image3Pipeline object and running the step. Each time the step finished successfully (success here means not crashing, not performing a successful alignment). |
|
Comment by John Scott on JIRA: There is a possibly related error in jw01537-o003_20240616t141518_image3_00004: Same part of the code, and other products in this observations have the malformed polygon error... 2024-06-16 17:01:43,594 - stpipe.Image3Pipeline.tweakreg - INFO - ***** tweakwcs.imalign.align_wcs() started on 2024-06-16 17:01:43.5945952024-06-16 17:01:43,594 - stpipe.Image3Pipeline.tweakreg - INFO - Version 0.8.62024-06-16 17:01:43,594 - stpipe.Image3Pipeline.tweakreg - INFO - 2024-06-16 17:01:44,551 - stpipe.Image3Pipeline.tweakreg - WARNING - Group with ID '987654' will not be aligned: empty source catalogTraceback (most recent call last):File "/dms/local/jwst/pipeline/pkgs/miniconda3/envs/jwstdp-1.14.0.20240401-py3.11/lib/python3.11/site-packages/stpipe/cli/strun.py", line 24, in mainStep.from_cmdline(sys.argv[1:])File "/dms/local/jwst/pipeline/pkgs/miniconda3/envs/jwstdp-1.14.0.20240401-py3.11/lib/python3.11/site-packages/stpipe/step.py", line 191, in from_cmdlinereturn cmdline.step_from_cmdline(args)^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^File "/dms/local/jwst/pipeline/pkgs/miniconda3/envs/jwstdp-1.14.0.20240401-py3.11/lib/python3.11/site-packages/stpipe/cmdline.py", line 394, in step_from_cmdlinestep.run(*positional)File "/dms/local/jwst/pipeline/pkgs/miniconda3/envs/jwstdp-1.14.0.20240401-py3.11/lib/python3.11/site-packages/stpipe/step.py", line 509, in runstep_result = self.process(*args)^^^^^^^^^^^^^^^^^^^File "/dms/local/jwst/pipeline/pkgs/miniconda3/envs/jwstdp-1.14.0.20240401-py3.11/lib/python3.11/site-packages/jwst/pipeline/calwebb_image3.py", line 89, in processinput_models = self.tweakreg(input_models)^^^^^^^^^^^^^^^^^^^^^^^^^^^File "/dms/local/jwst/pipeline/pkgs/miniconda3/envs/jwstdp-1.14.0.20240401-py3.11/lib/python3.11/site-packages/stpipe/step.py", line 509, in runstep_result = self.process(*args)^^^^^^^^^^^^^^^^^^^File "/dms/local/jwst/pipeline/pkgs/miniconda3/envs/jwstdp-1.14.0.20240401-py3.11/lib/python3.11/site-packages/jwst/tweakreg/tweakreg_step.py", line 476, in processalign_wcs(File "/dms/local/jwst/pipeline/pkgs/miniconda3/envs/jwstdp-1.14.0.20240401-py3.11/lib/python3.11/site-packages/tweakwcs/imalign.py", line 652, in align_wcsraise ValueError("Too few input images (or groups of images) with "ValueError: Too few input images (or groups of images) with non-empty catalogs. ```
ValueError: Too few input images (or groups of images) with non-empty catalogs
|
|
Comment by Mihai Cara on JIRA: should be fixed via spacetelescope/tweakwcs#205 and tweakwcs release 0.8.8 and #8657 |
Issue JP-3636 was created on JIRA by Hien Tran:
while reprocessing with b10.1.2, ops saw nis_image datasets for several programs crash in level 3 processing during the tweakreg step, as documented in DMSOPS-1090. an example is jw01536-o070_20240525t183345_image3_00002 (for product named jw01536-o070_t004_niriss_clearp-f444w-sub128)
The text was updated successfully, but these errors were encountered: