Following an updated version of PixInsight released yesterday, and a bug fix release today, data files produced by SGP are causing errors in the ImageIntegration process. The errors are due to the format of the SITELAT and SITELONG keyword valuess in the FITS headers of the files produced by SGP. The PixInsight ImageIntegration process generates errors, but completes and seems to generate correct results despite the errors.
Apparently previous versions of PixInsight ignored these FITS keywords in the ImageIntegration process, but now it is attempting to parse the keyword values as numbers, which produces the error messages. I posted about the error on the PixInsight forum, and the developer informed me that the format of the keywords is wrong in the data files produced by SGP. He cited the SBIG proposal for FITS keyword definitions, which states that the SITELAT and SITELONG keyword valuess should be in the format DD MM SS.
Here is a link to a document containing the proposed keyword definitions::
No @Marcelof, this problem was not fixed with today’s update. The issue with the RA keyword value was fixed, but the issue with the SITELAT and SITELONG keyword values was not fixed.
You will not see the errors if you are re-integrating files that are cached. Uncheck the “Use file cache” checkbox in the Image Integration parameters and you will see the errors again when you re-integrate. That’s assuming you have SITELAT and SITELONG keyword values in your FITS files. These values come from the User Profile settings in SGP.
FYI, there was another PI patch released today (1.08.06.1475) that fixed a different problem with parsing the SITELAT and SITELONG keyword values. The problem with parsing SGP-generated FITS files was not fixed. The PI developer has stated that this issue is caused by incorrectly formatted keyword values in FITS files generated by SGP, and he does not plan to provide a patch in PI. He also stated that the site latitude and longitude values are now being used in PI’s AnnotateImage script, and will be used in other PI processes in the future. So fixing this issue in SGP is necessary to maintain compatibility with PixInsight, which I suspect many of us use to process our images.
Here is a link to the PI forum thread in which I raised this issue and the PI developer responded. Scroll down to the third post in the thread:
I know this has been fixed since this thread was started, but what was the impact of this in overall image processing w/PI w/SGP? I have some old captures being processed now from 2019 where these images were caught in the crossfire here.
For example, would LocalNormalization in PI not work properly because PI can’t read/parse the SITELAT/LONG values?
