Error within StarAlignment

[micnek]

Hi friends!
I'm new to Pixinsight but did some astrophoto before.
I tried to process 7 exposures á 60 sec of M42 (+10 darks +10 bias) and it worked nicely but resulted in a monochrome picture. I read a tutorial again, and learn that you need to check the box "CFA images" to get color images, so I did that. However, the program crashed and was unable to deliver any images at all. I got the following error msg:
***
Batch preprocessing Script 1.46
StarAlignment
Cannot execute instance in the global context.
Reason: No target images have been defined.
***

What does this mean and how can I avoid this error?

I tried to uncheck the CFA box - and the process went well again - but resulted in b/w image of course.

Thank you very much for the expert opinion!

[sharkmelley]

You don't say which camera you are using.  Since you mention CFA, I'm assuming you are using a DSLR or one-shot-colour camera.

It is the Debayer process that "develops" the raw images i.e. creates a colour image from the monochrome raw data.  It must be run before StarAlignment. 

My StarAlignment process doesn't have a "CFA images" checkbox.  Do you mean the "Generate drizzle data" checkbox, which does something entirely different?

If you are not using a DSLR or OSC then ignore my comment.

Mark

[micnek]

Thank you Mark for your comment!
You're right - I use Nikon 810A (DSLR). How do I run that Debayer process? Is it some option inside Pixinsight, or should I run it before loading images to Batch Preprocess?
***
You can see which box "CFA images" I mean - see the screenshot in the attachment.

***
Have you any suggestion what does the error msg mean: "Cannot execute instance in the global context. Reason: No target images have been defined"?

Thx again for the assistance!
/Michael

[sharkmelley]

I was being dense - I hadn't spotted that you are running BPP 

BPP should do the debayering automatically if "CFA Images" is checked.

I don't know what's going on. Did you check the "Save process log"?  This will generate a log file.  Unless anyone else knows what's happening, I think we need to see that log file.  Hopefully it will tell us what's going wrong.

One other thing, are your Nikon files NEF files or are they FITS?  If they are FITS files then you need to manually select the Bayer/mosaic pattern instead of using Auto.  Possibly that is the problem.

Mark

[micnek]

Hi Mark!
I use NEF files, so it should work in auto-mode, right?

I tried to run Batch Preprocessing again, this time added few more frames (totally 7 frames á 60 sec + 5 frames á 120 sec). This time batch process did not crush and run through smoothly, but I still get monocrome images in the end.
I noticed one thing which differs compared to the tutorial: in the tutorial they use *.fit files for image integration, but I only get *.xisf files in the "registered" folder. Does this difference matter?
Thx!

[sharkmelley]

Yes NEF files should work fine in auto-mode.

Did you have "CFA images" checked for this latest successful run - if so, the registered images should not be monochrome.

XISF files are normal with the later versions of PixInsight.

I think we need to see a log file.

Mark

[micnek]

I run Batch preprocessing again, and this time it crushed as described earlier. So far, I only succeeded to process images sometimes (with b/w result), but mostly no result at all.
Here comes my process log as pdf. Is there a better way to export a log?
/M

[micnek]

When I look as the log file myself, it seems to be a problem with some of the Darks are taken with 300mm lens, and some are taken with 135 mm lens. I did not know that was important (thought that temperature and exposure time were important, but not the lens). I will try again with more consistent Darks.

[micnek]

Well, I tried again, with 100% consistent images (same exposure, ISO, lens in both darks and lights), but batch preprocess still crushed. Here is my log - it seems to be an error in aquiring CFA-image data ("invalid image properties"), but I don't understand why.

[micnek]

The last log seems not to be complete - here we have one more failed Batch Preprocess...

[sharkmelley]

The log file was very helpful.  The relevant part is where the raw files are decoded:

Invoking: dcraw -4 -o 0 -r 1 1 1 1 -t 0 -k 0 -H 1
Decoding Nikon D810A file (10548439682656x2456 pixels, ISO=400, Exposure=1/2500s): done
Loading raw image: done
Computing image statistics: done
Weight : 1.00000
/Volumes/LaCie/Astro/Darks and Flats D810A/Bias/_DSC0094_ISO400.NEF
Timestamp: Sun Dec 10 00:36:52 2017
Camera: Nikon D810A
ISO speed: 400
Shutter: 1/2500.0 sec
Aperture: f/4.0
Focal length: 300.0 mm
Embedded ICC profile: no
Number of raw images: 1
Thumb size: 3680 x 2456
Full size: 3680 x 2456
Image size: 3680 x 2456
Output size: 3680 x 2456
Raw colors: 3
Daylight multipliers: 2.062443 0.937244 1.186358
Camera multipliers: 2.183594 1.000000 1.171875 0.000000


I've highlighted the bits that seem suspicious - are your files really that small - only 9 Mpixels?
In addition, PixInsight hasn't picked up the filter pattern - it should normally report something like this:
"Filter pattern: RG/GB"

The lack of filter pattern i.e. the CFA pattern is what is causing your issue with the Debayer process

So I'm wondering if you have shot your data is some strange shooting mode?  Or maybe PixInsight cannot interpret D810A files.  The fact that it reports "10548439682656x2456 pixels" is pretty weird.

You might be able to process the files by explicitly specifying RGGB as the CFA pattern.  But I'm still suspicious that it is reporting an image size of 3680 x 2456.  What size is the image if you open it in any other image editor?

Mark

[micnek]

Thanks a lot for your time solving my tiny problems!
It's actually getting more interesting.

1. On Nikon 810A (btw letter "A" means "Astronomy version"), there is no filter installed before the sensor - to let wider wavelengh spectrum through.
2. I have an option to choose between large raw files 7360 x 4912 and small raw files 3680 x 2456. I normally use small raw, especially for astrophoto. I will try to process large raw, but this probably is not the reason for trouble.
3. I did nothing unusual when shooting - all light frames and darks are taken with 60s exposure, bias frames are taken with 1/2000 or something like that.

So, it looks like a genuine bug in the software, which I did not believe from start (more likely it would be my own "human error")?
Maybe, Pixinsight does not recognize images as color, since it can not detect any filter??
- Should we report this to someone? I mean, it's unfortunate if Nikon's only astro-camera is not compatible with Pixinsight?
- What shall I do so far? Is it possible to split color image in three RGB channels, process them separately and then put together again? A bit cumbersome, but might work?

[sharkmelley]

It might be the small format causing the problem somehow. Can you upload one of your lights - in NEF format so I can take a look and better understand what may be happening.

Mark

[micnek]

Hi, Mark! I sent one of my NEF-images to your e-mail adress which is on your website. The file is 27 MB, should be OK to mail it. Oherwise I can send it by e.g. sprend.
Thanks a lot for your help!

[micnek]

Hi again! It seems that you solved the problem, Mark! I tried to process some old images of M51, which was taken with full resolution (large raw), and it worked!
I feel like Christmas now - thanks a lot! ))