Missing FITS Info Question

Reggie Jones

Hi.  I may have asked this previously but I'll ask again.  Was using PCC to color calibrate and image and it would not plate solve.  Tried to use Image Solver and got the same error.  I checked the FITS header info in the combined RGB file after Channel Combination and there was no information.   There was also no FITS header info in the R,G,B Masters.  I looked at an individual Luminance sub exposure which did have FITS info, I took OBJECT RA and OBJECT DEC info from that file, used in PCC and everything worked fine.   Still puzzled as to why the FITS info would be missing from the Integrated Master R,G,B and combined files (done thru Channel Combination); it's the second time I've run into this issue.  Have you seen this problem?  Wondering if I missed doing something in pre processing.

Thanks

BlockHead

Keep on eye on cloned images (or copies of previews). I am pretty sure FITs info does not carry forward form them. You should be able to determine at which point the FITs header goes away regardless of the above idea.

-the Blockhead

Reggie Jones

Ok, Thanks

Reggie Jones

FYI - backtracked and discovered that when I used Normalized Scale Gradient on the data after Star Alignment, the FITS Header was changed and the information was removed from the data.  

Reggie Jones

FYI - figured out what I did wrong.  Somehow, I'd inadvertently checked the "Overwrite Existing Files" when I ran the NSG script.  That destroyed the original data in the FITS header as is advertised.

BlockHead

Ah ha! Interesting. 

-the Blockhead

BlockHead

Hi Reggie,

So... I connected with John Murphy. You might have discovered a bug. But he can't figure out how it happens exactly. The condition you describe can happen if the data was in the form of integers instead of floats. 

Was were the images you used XISF files that were outputted from StarAlignment? ( assume so... )

-the Blockhead

Reggie Jones

Yes. And I checked the files after StarAlignment and the FITS info was still there. It was only after using Normalized Scale Gradient that the FITS info came up missing.

John Murphy

If you can provide me with a reference and target image that demonstrate the problem, I will take a look.

Regards, John Murphy

Reggie Jones

Hi John; sorry its taken me a bit to respond.  If you send me an email address, I will share the DropBox folder I share with Adam that contain 2 XISF files.  The registered file is output from the Star Alignment Process which still has all the FITS Header info.  The NSG file is the same registered file but now output from the Normalized Scale Gradient script and there is no FITS Header info.

Thanks

Reggie Jones

John Murphy

I have sent a message that contains my email address.

Regards, John

Reggie Jones

Got it and shared the folder.  Let me know if you need anything else.

Reggie

John Murphy

Hi Reggie

The source .xisf file was in Integer 16 format.

NSG reads in the image and then has to convert its internal copy of the image to floating point. There was a bug in this conversion - I had forgotten to copy over the FITS headers.

This bug was fixed yesterday. See PixInsight forum post:

https://pixinsight.com/forum/index.php?threads/normalizescalegradient-1-4-6-beta-rescales-images-to-avoid-truncation.16507/post-108010

Please try it with this new version and let me know if it works.

Thanks,

John Murphy

Reggie Jones

Thanks John.  So, I'm was working on a different dataset today and followed my normal workflow with NSG; I checked the NSG files and the FITS information did come across as expected.  Looks like it all works.

Thanks for taking the time to help resolve this.

Reggie

BlockHead

Wait... how was this file an integer file??

That is still amystery?

-the Blockhead

Reggie Jones

Hi Adam. I’m not sure I can answer that one. I’m not that well versed in file structure.

Reggie

John Murphy

I think it is possible that this data was not calibrated by PixInsight. Was the image from TelescopeLive? Are these images calibrated by TelescopeLive?

The first evidence of processing that I see in the FITS header is CosmeticCorrection (PixInsight 1.8.8-11)

I think this might explains why the data has not been converted to floating point.

If this is the case, the Integer - missing FITS header bug may hit many people. I will try to get this bug fix included in an update, rather than waiting for the next release.

Regards, John Murphy

Reggie Jones

Hi John - this was Telescope Live data.  Specifically from their 1Click bundled data of NGC 4372, the Dark Doodad nebula.  The data I get from them, whether it's an Advanced Request I made or like this the data is always calibrated by them.  

After I download and Blink through it, unless there is something really unusual with it the next step I take usually is with Cosmetic Correction.

Reggie

BlockHead

I see.

But I am still curious. The data must be registered using StarAlignment. The interpolation I would have though to result in floating point results?

-the Blockhead