Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Problem with debayer Canon EOS images

  • Please log in to reply
11 replies to this topic

#1 konstantinos

konstantinos

    Vostok 1

  • -----
  • topic starter
  • Posts: 116
  • Joined: 01 Aug 2014

Posted 18 November 2015 - 08:52 AM

Hi

I use TheSkyx Pro to connect my Canon EOS 450D camera. The data acquisition mode I have chosen is "Raw Bayer Data".
I have told TheSkyx to save the images as FIT format.

This is the information provided by TheSkyx about this mode

1. Raw Bayer Data - Only images shot in this mode can be converted to color later. This mode returns the raw Bayer matrix data from the camera. Raw Bayer data contains four pixels cells that contain one red, one blue, and two green pixels (laid out in RGGB order). The image returned to TheSkyX is in the native resolution of the camera. Image link works quite well on this data despite that it appears to be missing pixels (the blue pixels tend to be very dark). When magnified you will see a characteristic screen door type pattern. This data does not show in color. A post processing program is required (PixInsight, Nebulosity, etc.) to convert the Bayer data to a color image. Best results are obtained by doing image calibration (darks and flats) on this raw pattern before conversion to color. You must also convert to color BEFORE image alignment and stacking operations, or the color data will be lost.

Now you can bin 2x2 while in Raw Bayer mode, which is effectively a summed luminance mode (you CANNOT recover color data if you bin)

Then I pass a FIT image to Pixinsight. I make image calibration then I proceed to debayer the image so that color will be shown.

The problem I get is that on the star halos I see some pink/red ringing side effects.

I have tried all debayering combinations in Pixinsight and the problem is always there.


Does anybody has experienced this problem and has some solution to it other than process further this image e.g. in Photoshop play with selective colors to get rid of the artifacts?

Thank you

Attached Thumbnails

  • problem.png
  • problem_maximized.png


#2 DrewS

DrewS

    Sputnik

  • *****
  • Posts: 44
  • Joined: 23 Jan 2008

Posted 18 November 2015 - 02:40 PM

If you look at a single image (before stacking or calibrating) with something like the Canon Digital Photo Professional viewer or similar) do you still see the color fringes? Or if you save as RAW+Jpg do you see the fringe in the Jpg?  They may be real i.e. slight chromatic aberration in your optics, not any artifact of de-bayering.

 

What scope or camera lens are you using?

 

Drew S.



#3 konstantinos

konstantinos

    Vostok 1

  • -----
  • topic starter
  • Posts: 116
  • Joined: 01 Aug 2014

Posted 18 November 2015 - 04:16 PM

Hi

 

I will need to repeat this and check the following

 

(a) accurate focus is performed which will eliminate any color fringes on the stars

(b) let TheSkyx save the images as FIT and RAW format

© debayer the FIT and see if problem appears again

     check the RAW and see if problem appears

 

The camera lens is Canon EF 200mm f/2.8 L

 

Thank you



#4 rigel123

rigel123

    ISS

  • *****
  • Posts: 20,446
  • Joined: 29 Jun 2009
  • Loc: SW Ohio

Posted 18 November 2015 - 04:30 PM

Definitely not an expert on this but I would guess because of the lens you are dealing with chromatic aberration as mentioned.  In Light Room there is a function for reducing chromatic aberration which I use in my terrestrial shots with my canon.  You might also break out the individual channels as grayscale and then use software that will register the stars and then align the channels to help reduce the chromatic aberration.  You can also reduce the size of the stars in the channels that are showing bloated stars before combining to help reduce these halos as well.



#5 DrewS

DrewS

    Sputnik

  • *****
  • Posts: 44
  • Joined: 23 Jan 2008

Posted 19 November 2015 - 12:28 AM

Unfortunately accurate focusing won't remove chromatic aberration (the red doesn't focus in the same place as the blue because of prism effect and nudging the focus won't alter that)

 

A camera lens will very very likely have chromatic aberration (and a lot of telescopes will too).

 

The shift of one color relative to another will likely vary as the distance from the center of the lens (which means you can't just nudge the red channel a little to match the blue).

 

As Rigel123 suggest reducing the size of the star images in each channel may help. You probably can't fix this in processing, though.

 

Drew S.



#6 konstantinos

konstantinos

    Vostok 1

  • -----
  • topic starter
  • Posts: 116
  • Joined: 01 Aug 2014

Posted 29 November 2015 - 12:30 PM

Hi

This time I have told TheSkyx to save the images as RAW format also.
When I load the RAW image in Pixinsight and debayer it afterwards, the problem no longer appears.
So there's must be an issue with TheSkyx saving Canon EOS 450d images as FIT format.
I cannot explain this other way...

Attached Thumbnails

  • test.jpg


#7 konstantinos

konstantinos

    Vostok 1

  • -----
  • topic starter
  • Posts: 116
  • Joined: 01 Aug 2014

Posted 02 December 2015 - 04:49 PM

Hi

 

Does anybody know how to get rid of these red/magenta color fringes on the stars with the help of Pixinsight?

 

I will try to split the image in R,G,B channels and reduce the stars in R channel, then recombine the RGB.

Any other ideas are welcomed.

 

Here you can download the fit file: https://www.dropbox....D_crop.fit?dl=0

 

Thank you

Attached Thumbnails

  • problem.jpg

Edited by konstantinos, 03 December 2015 - 07:26 AM.


#8 tommy_h

tommy_h

    Ranger 4

  • -----
  • Posts: 334
  • Joined: 27 Dec 2011
  • Loc: Zellerndorf/Austria

Posted 03 December 2015 - 08:33 AM

Modify the settings of the pixel rejection (sigma high, sigma low) in the ImageIntegration process!

 

Thomas



#9 David Ault

David Ault

    Gemini

  • -----
  • Posts: 3,424
  • Joined: 25 Sep 2010
  • Loc: Georgetown, TX

Posted 05 December 2015 - 01:57 PM

Hey Konstantinos,

 

There's a couple threads about how to improve this using a bayer matrix based drizzle algorithm in PI.  This is the first one I remember.  Someone else came up with a script to do some of the intermediate processing steps and called the script BayerDrizzlePrep.  I can't guarantee it will fix this specific problem as there are multiple causes for issues like this, but it would be good to try and see if it helps.

 

Regards,

David



#10 Michael Covington

Michael Covington

    Author

  • *****
  • Posts: 6,953
  • Joined: 13 May 2014
  • Loc: Athens, Georgia, USA

Posted 05 December 2015 - 02:32 PM

Hi

This time I have told TheSkyx to save the images as RAW format also.
When I load the RAW image in Pixinsight and debayer it afterwards, the problem no longer appears.
So there's must be an issue with TheSkyx saving Canon EOS 450d images as FIT format.
I cannot explain this other way...

 

There are different deBayering algorithms and it sounds like PixInsight has a better one!



#11 konstantinos

konstantinos

    Vostok 1

  • -----
  • topic starter
  • Posts: 116
  • Joined: 01 Aug 2014

Posted 05 December 2015 - 05:37 PM

Hi

 

My problem is solved, or at least now the stars look better than before, thanks to David suggestion about Pixinsight new functionality.

 

I have followed these steps in Pixinsight

 

 

Step 1. Calibration of the FIT files generated by Skyx, with bias/darks/flat frames. Cosmetic correction also performed.
Step 2. Debayering of the calibrated files using RGGB pattern and VNG algorithm
Step 3. Registration of the debayered files and generation of drizzle files
Step 4. Integration of the registered files and generation of new drizzle files
Step 5. Invoke BayerDrizzlePrep script in Pixinsight

             Images to convert=calibrated images of step1 above

             Drizzle files to convert=drizzle files generated from step4 above 

Step 6.  Drizzle integration of the the new drizzle files generated from step5 above.

 

I attach an image for your reference to compare the before and after.

 

Thank you all for your contribution to this difficult topic

 

Konstantinos

 

 

 

 

Attached Thumbnails

  • comp.jpg


#12 DuongDT

DuongDT

    Lift Off

  • -----
  • Posts: 7
  • Joined: 03 May 2018

Posted 20 May 2018 - 10:52 PM

Sorry for digging up this thread! I have a problem with preprocessing and processing images of M8 and M20/21 taken with Canon 550D and the SkyX version 10 by PixInSight. After making master dark, master bias, I calibrated the light frames with master dark and master bias. The calibrated frames are then debayered using RGGB pattern and integrated into the "final" image for processing. The image has green tint as usual. However, when I stretched the image, carried out Histogram Transformation and Dynamịc Background Extraction, the colors show up very poor: M20 is orange and M8 has a pink-ish tint. I don't know why. I have checked the FIT header, it shows RGGB pattern

I have been processing Canon's CR2 raw format with great success for 2 years, but I'm struggling with fit images taken by The SkyX.

I am looking forward to your advice. Thank you 




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics