Observatory 2.0: PixInsight 1.8.9-2 astrometrical data not recognized

[avalondria2050]

Hi.

I have imported several new images that includes astrometrical data from the latest PixInsight version ( 1.8.9-2 ) to Observatory 2.0. I haven't done this for a while, but the last time I imported and image from PI to Observatory 1.6.x it was able to show the astrometrical data ( I had to play with the image orientation in order to show it right, but it worked with this workaround ). Now in Observatory 2.0 it just says: "This image does not contain an astrometrical solution", but these images have been solved correctly in PI.

Any idea why this could be happening?

Thank you

[Sander Berents]

I typically wait a few weeks before I update my copy of PixInsight so I didn't yet. When this one was announced I noticed a big red flag in the release notes. According to those this latest version doesn't write the WCS solution as FITS headers anymore!

From https://pixinsight.com/forum/index.php?threads/pixinsight-1-8-9-2-released.21228/:

Our astrometric solutions are now based exclusively on XISF properties and can only be stored in XISF files. We no longer depend on FITS header keywords and don't generate any WCS FITS keywords.
...
However, all WCS keywords are removed from the image's metadata, and a new astrometric solution is generated automatically as a set of XISF properties following our coordinate system conventions and implementing features unique to our platform.


I didn't check if these coordinate system properties are documented in the XISF documentation. The reason for dropping this is kind of weak in my opinion and smells like vendor lock-in. Especially if these are not even written in FITS files anymore. I do intend to look closer into these XISF properties though and hopefully can work around this unnecessary change.

[avalondria2050]

Yes, I guess they are trying to move everything to .XISF files to lock the customers to their solution. Hope this can be solved, because I tried to solve the images with the internal Observatory 2.0 and it struggles with images full of stars ( like NGC7000 or similars ).

Best Regards

[Sander Berents]

If PixInsight lost the capability to write WCS headers to FITS files, which is how I interpret the release notes, then that's a real step backwards. Not writing them in XISF files but use XISF properties instead is fine. It's PixInsight's own image format after all. Anyway, I will look into this.

Could you please temporarily share one of your images (e.g. using iCloud) so I can investigate the issue you are having with the new plate solver? I find it way more robust than the old one, but it always can get better.

[avalondria2050]

Sorry my delay. Here you are an example of a starry area that Observatory can not solve ( I have tried but it refuses to ). When there is an isolated object, like a galaxy or a planetary nebula, it solves the image correctly. But with so many stars the process fails . Try yourself. I am able to solve it with PI, but for now the metadata can't be shown.

I have uploaded the image to Google Drive, as it's bigger than 192KB.

https://drive.google.com/file/d/1io7qzhoQo82sGPzN-Pv7yFwI8_9_Vjbn/view?usp=sharing

[Sander Berents]

I think you didn't share it correctly. It asks for a request to access it, which I sent yesterday, but you never responded. Please share the file(s) for public access so I can download it. Don't share your whole Google Drive of course. After that remove the share again. Or share with iCloud.

[avalondria2050]

Yes, you are right. I don't use Google Drive too much. I have granted you the access, so you should be able to download the file now.

[Sander Berents]

Thanks, I downloaded it. You can remove the share now. My God, It's Full of Stars

[avalondria2050]

Yes, it is :-P

[Sander Berents]

Observatory 2.0.2 supports linear astrometric solution XISF properties, so it and its Quick Look preview extension is able to render the overlays for XISF files written by PixInsight 1.8.9-2.

[avalondria2050]

Hi.

Yes, I have seen that now Observatory can use the latest PixInsight astrometric data. Thank you for implement it. The next "frontier" is to make Observatory to deal correctly with the "megastarry" images when solving internally.

Best