Jump to content


Photo

DSLR Debayering- criss cross patterns?

  • Please log in to reply
10 replies to this topic

#1 CMS1288

CMS1288

    Explorer 1

  • -----
  • Posts: 82
  • Joined: 01 Apr 2013
  • Loc: Long Island, NY

Posted 06 April 2013 - 11:09 PM

Hello everyone, this is my first post!

I've been taking my first astro images and quickly learning all the bits and pieces. I have an unmodified Nikon D3100.

My question is, I batch debayer in Nebulosity and pre-process as usual, then I finish post processing there and save a 16-bit Tiff and use PSE11 to finish up. However, if I start zooming in on the image, a grid pattern shows and changes as i keep zooming in. I'm assuming it's the Bayer matrix right? Any calibration settings that may alleviate this?

Thanks!

#2 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4869
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 06 April 2013 - 11:19 PM

Welcome to Cloudy Nights CMS1288!

Curious about the grid pattern. Does it show up in Nebulosity as well or only in PSE? Can you post a 100% resolution crop of the image?

As far as I am aware Nebulosity does not do a convert-to-colour by simply converting each pixel of the bayer matrix.... the batch debayer SHOULD have put colour data from all three channels in each pixel location (and thus get rid of being able to see the bayer matrix itself).

#3 zerro1

zerro1

    Fly Me to the Moon

  • *****
  • Posts: 5850
  • Joined: 02 Aug 2009
  • Loc: Smokey Point , 48.12°N 122.25°W Elevation:512 ft

Posted 06 April 2013 - 11:22 PM

Debayering shouldn't be done until after calibration, and just prior to alignment and stack. The bayer pattern shouldn't be present after "demosaic". though if one gets too carried away zooming in on the image it will become "pixelated"

#4 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4869
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 06 April 2013 - 11:27 PM

Oh, ya, I missed that. If you batch debayer the LIGHTS and then calibrate with un-debayered darks/flats/etc it would induce some funkyness.

As Robert said the process should be:

- Pre-process (Darks, Flats, Flat-darks)
- Batch Debayer
- Align and Combine (stack)
- Initial post-processing
- Save as 16bit TIFF
- final post-processing in PSE

Still - seeing an example of one of your files showing the grid/criss cross pattern would be helpful to confirm.

#5 Tonk

Tonk

    Fly Me to the Moon

  • *****
  • Posts: 7078
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 07 April 2013 - 05:14 AM

Any calibration settings that may alleviate this?


If its not the above explanations - you could be seeing the sensor bias pattern, in which case shoot some bias images and throw those into the calibration mix

Also if you are adding flats you MUST shoot dark flats or bias images for those flats. Not calibratring your flats correctly is another route to the tartan stripes

#6 CMS1288

CMS1288

    Explorer 1

  • -----
  • Posts: 82
  • Joined: 01 Apr 2013
  • Loc: Long Island, NY

Posted 07 April 2013 - 08:07 AM

Thanks for the replies....

Yes I'm doing the batch demosaic after pre=processing and normalizing, and just before aligning and stacking. I'm also calibrating with darks and flat frames (calibrated with bias).

Oh, and it only shows up in PSE. On Nebulosity, it's fine.

Attached Files



#7 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4869
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 07 April 2013 - 11:27 AM

ok that is strange... so when you zoom in to 200% or 400% in Nebulosity you do *not* see this pattern? What version of Nebulosity are you running? Have you tried opening the 16bit TIFF in programs other then PSE?

#8 CMS1288

CMS1288

    Explorer 1

  • -----
  • Posts: 82
  • Joined: 01 Apr 2013
  • Loc: Long Island, NY

Posted 07 April 2013 - 10:36 PM

I'm running Nebulosity 3.1, the current version. I'm running on mac, and the preview application shows the same issue. I tried again with Nebulosity, and the pattern DOES exist, just not as extreme- but it's there.

This pattern looks worse at certain zooms; at times it looks like a Scottish kilt with big squares or red and green.

This is the only image I've found that resembles it (scroll down to debayering): http://www.sergepetiot.com/?page_id=22

Seems like a debayering issue to me, but I'm no expert. I haven't touched the settings box that comes up in Nebulosity when one debayers.... are there certain settings I should be using?

#9 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4869
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 07 April 2013 - 11:14 PM

Well that does indeed seem to be a debayer issue...

Could you post a RAW file from your camera online someplace? I also have Nebulosity 3.1 and run on Mac - I would be interested to see if I could replicate the issue. In fact if you can post at least one each Light, Dark, flat, & Bias frame it would help to determine the source of the problem should it not be debayer settings.

#10 CMS1288

CMS1288

    Explorer 1

  • -----
  • Posts: 82
  • Joined: 01 Apr 2013
  • Loc: Long Island, NY

Posted 08 April 2013 - 08:40 AM

Well it looks like I figured it out... for some reason, my X and Y matrix offsets were both 0. I moved the X offset to 1 and it solved everything- the criss crossing, the star colors, and the colors of the galaxy all fixed.

#11 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4869
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 08 April 2013 - 12:23 PM

Glad you where able to resolve the issue. I guess Nebulosity did not have a pre-programed profile for the bayer matrix in that particular camera (for cameras it *knows* it does not ask about the offset).

Hope we will see the end result posted here sometime soon!






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics