DATE-OBS calculation seems inconsistent

[txastro]

Hello,
I do not know if this is a bug.  I have 109 subs for comet 2016/R2 PanSTARRS.  After loading them into CometAlignment they are not in order by datetime taken.  Sub #96 is taken as the first image and #95 as the last image.  I've attached screenshots of FITSHeader for those two images.  For #95 the value for DATE-OBS is correct (US central standard time + 6).  For #96 however, DATE-OBS shows the time as US central standard time + 0.  Camera used is a Canon T6i and CR2 to XISF done in PixInsight only.  The CR2 filename was created using EXIFTool to include date and time.  Frames after #96 all have DATE-OBS as central standard time + 0.  So, they are at the top of the list.

Thanks.
Xavier

[Juan Conejero]

Hi Xavier,

This is not a bug in PixInsight. Please note that PixInsight cannot generate DATE-OBS FITS keywords, unless you have acquired the frames with our INDICCDFrame tool. In any other case our tools just copy the existing DATE-OBS FITS keywords (among many others) to newly created files without modifying them in any way. If you have acquired the frames with other applications, then the problem has been caused by those applications.

[txastro]

Hi Juan,

Good info to know.  The CR2 frames are in-camera, no external capture application.  I thought that the FITS keyword was populated with the timestamp created by the T6I when Pixinsight calibrated the CR2 light frames and created the XISF files.  I appreciate the response.

Thanks,
Xavier

[Juan Conejero]

Hi Xavier,

I thought that the FITS keyword was populated with the timestamp created by the T6I when Pixinsight calibrated the CR2 light frames and created the XISF files.

That's true; I forgot about this feature. However, PixInsight just copies the timestamp included in the raw frame, normally as EXIF metadata, and generates a DATE-OBS keyword with an ISO 8601 representation of the timestamp. The existing value should not be modified in any way.

[txastro]

Hi Juan,
Maybe there is something happening when the T6i creates the timestamp, perhaps a Canon issue with representing UTC.

Thanks!
Xavier