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

Binning One Shot Color (OSC) Cameras

  • Please log in to reply
9 replies to this topic

#1 stevel3o7

stevel3o7

    Lift Off

  • -----
  • topic starter
  • Posts: 23
  • Joined: 30 Aug 2017
  • Loc: Hollister CA

Posted 13 April 2021 - 08:53 PM

I have a ZWO ASI2600mc pro, one shot color (OSC), 3.76u pixels, CMOS.

 

This camera on an SCT with a .63 focal reducer gives .58/px resolution which is oversampled.  Using 2x2 binning the resolution increases to 1.14/px which is fine.  However I want to retain the color information, and using 2x2 binning on the CMOS chip color information is lost.

 

Have noted threads on CN that mention it is better to bin in post; what software does this?  I have DSS, Registax, Sequator etc.  

 

Is software binning the same as resampling?

 

Using the 2x2 ZWO mono bin lowers the resolution, increases images scale and increase signal to noise (but not as much as with mono camera) but there is loss of color information (which is lost with any in-camera binning on CMOS.)  

 

Was thinking of shooting color at the oversampled rate, and then shoot using the mono 2x2 binning, and then combining the images in photoshop using layers, similar to LRGB imaging.   Curious if anyone has tried this?

 

Thank you



#2 Midnight Dan

Midnight Dan

    Voyager 1

  • *****
  • Posts: 14,513
  • Joined: 23 Jan 2008
  • Loc: Hilton, NY, Yellow Zone (Bortle 4.5)

Posted 13 April 2021 - 09:01 PM

I have a ZWO ASI2600mc pro, one shot color (OSC), 3.76u pixels, CMOS.

 

This camera on an SCT with a .63 focal reducer gives .58/px resolution which is oversampled.  Using 2x2 binning the resolution increases to 1.14/px which is fine.  However I want to retain the color information, and using 2x2 binning on the CMOS chip color information is lost.

 

Have noted threads on CN that mention it is better to bin in post; what software does this?  I have DSS, Registax, Sequator etc.  

 

Is software binning the same as resampling?

 

Using the 2x2 ZWO mono bin lowers the resolution, increases images scale and increase signal to noise (but not as much as with mono camera) but there is loss of color information (which is lost with any in-camera binning on CMOS.)  

 

Was thinking of shooting color at the oversampled rate, and then shoot using the mono 2x2 binning, and then combining the images in photoshop using layers, similar to LRGB imaging.   Curious if anyone has tried this?

 

Thank you

Resampling is the same as binning.  But there are different algorithms that can be used to resample, so you want to use the right one.  Normally for binning, you just want to average the 4 pixels.  Pixinsight has a process called IntegerResample that does this.  Not sure what the equivalent would be in your processing software.  But you want to bin *after* you have done all your calibration, registration, and stacking.

 

-Dan


  • PirateMike likes this

#3 jdupton

jdupton

    Gemini

  • *****
  • Posts: 3,372
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 13 April 2021 - 09:09 PM

stevel3o7,

 

   The easiest possible way to bin an OSC camera is to choose "SuperPixel" as the deBayer mode during pre-processing. I think most astro software packages should support SuperPixel deBayering.

 

   What this method does is to simply take the Red Pixels and move them to the Red Channel of the deBayered image. It then takes the two Green Pixels and averages them then moves them to the Green Channel of the output image. Similarly, it takes the Blue Pixels and moves them to the Blue Channel of the output image.

 

   This operation has the natural side effect of reducing the image dimensions by a factor of two since each 2x2 group of pixels in the original exposure had one Red, Two Green, and one Blue pixels. The 2x2 group simply becomes a single three color channel pixel in the output. It is the cleanest way to deBayer in that no interpolation is required and it bins the image just as you wanted.

 

 

John


  • PirateMike and jerahian like this

#4 PirateMike

PirateMike

    Soyuz

  • *****
  • Posts: 3,843
  • Joined: 27 Sep 2013
  • Loc: A Green Dot On A Blue Sea

Posted 13 April 2021 - 09:23 PM

stevel3o7,

 

   The easiest possible way to bin an OSC camera is to choose "SuperPixel" as the deBayer mode during pre-processing. I think most astro software packages should support SuperPixel deBayering.

 

   What this method does is to simply take the Red Pixels and move them to the Red Channel of the deBayered image. It then takes the two Green Pixels and averages them then moves them to the Green Channel of the output image. Similarly, it takes the Blue Pixels and moves them to the Blue Channel of the output image.

 

   This operation has the natural side effect of reducing the image dimensions by a factor of two since each 2x2 group of pixels in the original exposure had one Red, Two Green, and one Blue pixels. The 2x2 group simply becomes a single three color channel pixel in the output. It is the cleanest way to deBayer in that no interpolation is required and it bins the image just as you wanted.

 

 

John

This is great information. I don't shoot OSC but I know some who do and sometime I am asked to process their images.

 

I'll give it a try next time I find some OSC frames have been emailed to me. wink.gif

 

Thanks John. waytogo.gif

 

 

Miguel   8-)

 

.


Edited by PirateMike, 13 April 2021 - 09:25 PM.


#5 Dynan

Dynan

    Fly Me to the Moon

  • *****
  • Posts: 5,712
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 13 April 2021 - 09:44 PM

Star Tools does post calibration binning.



#6 Jinux

Jinux

    Messenger

  • *****
  • Posts: 432
  • Joined: 25 Jun 2007
  • Loc: Bay Area, CA

Posted 13 April 2021 - 10:27 PM

John's answer above is right. 

 

Somehow, in astro community, there is a "perception" that "over"sampling is a bad thing. In signal processing world, if there is no other signal processing gain, 2x oversampling means 3dB SNR improvement. There is simply nothing wrong on oversampling. We don't do it simply because making fast ADC while preserving the same resolution is expensive. "over" is simply indicating term that it's more than Nyquist rate. If a system requires high resolution but can't make high resolution ADC, system designer often orders to make over-sampling ADC, so that they can do signal processing and get better SNR. In astro world, we do sample 'spatial' frequency with pixel array. 

 

I think misperception is coming from the two legacy technology facts and seeing/optics limitation. 

When you're seeing and/or optics resolution limited, having finer image scale doesn't help the final image much. So it doesn't make sense of investing on finer pixel camera when it was expensive to make one, and if it doesn't help the final image much.

In CCD technology, binning happens on sensor before ADC, so it could reduce read noise. Also, CMOS front-illuminated sensor has 'aperture' ratio which goes low as the pixel size gets smaller, so too small pixel was a bad thing. 

In modern back-illuminated CMOS sensor, these two are not true anymore and binning happens on driver layer by just adding values, so it doesn't help at all but just blocks any signal processing possibility like using SuperPixel DeBayer as mentioned above.

 

-Jinux



#7 TelescopeGreg

TelescopeGreg

    Gemini

  • -----
  • Posts: 3,111
  • Joined: 16 Jul 2018
  • Loc: Auburn, California, USA

Posted 13 April 2021 - 11:52 PM

On-camera 2x2 binning with the ASI2600MC camera does NOT result in a colorless image.  This was taken last October, 2020.  27 x 240 seconds, 2x2 bin in the camera, stacked with DSS, processed in StarTools.

 

I'm currently experimenting with 1x1 vs 2x2 binning with my setup.  1x1 has an 0.85" pixel scale, which is fine.  2x2 is 1.7", which is a bit big, but not out of the question, especially with typical seeing.  The resulting images aren't much different, as far as I can tell.

 

What I can tell you is that 1x1 binning fills up your SD card much faster than 2x2.  Don't ask me why I know that.

Attached Thumbnails

  • NGC7023 Iris Nebula 2 nights 1hr 48 min 240sec gain 300 ST new process.jpeg

Edited by TelescopeGreg, 13 April 2021 - 11:56 PM.


#8 mackiedlm

mackiedlm

    Messenger

  • -----
  • Posts: 421
  • Joined: 16 Jul 2019
  • Loc: Galway, Ireland

Posted 14 April 2021 - 03:00 AM

On-camera 2x2 binning with the ASI2600MC camera does NOT result in a colorless image. This was taken last October, 2020. 27 x 240 seconds, 2x2 bin in the camera, stacked with DSS, processed in StarTools.

I'm currently experimenting with 1x1 vs 2x2 binning with my setup. 1x1 has an 0.85" pixel scale, which is fine. 2x2 is 1.7", which is a bit big, but not out of the question, especially with typical seeing. The resulting images aren't much different, as far as I can tell.

What I can tell you is that 1x1 binning fills up your SD card much faster than 2x2. Don't ask me why I know that.


Interested to see the results of your experiment. Are you seeing better SNR with the binning?

#9 Midnight Dan

Midnight Dan

    Voyager 1

  • *****
  • Posts: 14,513
  • Joined: 23 Jan 2008
  • Loc: Hilton, NY, Yellow Zone (Bortle 4.5)

Posted 14 April 2021 - 08:39 AM

On-camera 2x2 binning with the ASI2600MC camera does NOT result in a colorless image.  

Correct.  Some ZWO cameras do support in-camera color binning.  However, look at section 4 in ZWO's page on binning:

 

https://astronomy-im...ndamentals.html

 

You'll note that they recommend using software binning instead if you want the benefits of binning.  Their in-camera binning just uses pixel skipping which means you'r throwing out much of the data.  They only provide that feature to allow for faster frame rates when that's the main concern.  If you're looking for the best resolution and the lowest noise, you need to use software binning.

 

-Dan


  • Dynan likes this

#10 TelescopeGreg

TelescopeGreg

    Gemini

  • -----
  • Posts: 3,111
  • Joined: 16 Jul 2018
  • Loc: Auburn, California, USA

Posted 14 April 2021 - 03:08 PM

Correct.  Some ZWO cameras do support in-camera color binning.  However, look at section 4 in ZWO's page on binning:

 

https://astronomy-im...ndamentals.html

 

You'll note that they recommend using software binning instead if you want the benefits of binning.  Their in-camera binning just uses pixel skipping which means you'r throwing out much of the data.  They only provide that feature to allow for faster frame rates when that's the main concern.  If you're looking for the best resolution and the lowest noise, you need to use software binning.

 

-Dan

Hmmm....  I read a different article on their site earlier, and came to the conclusion that they were averaging the pixels on-camera.  That the "apparent" download speed is significantly faster with 2x2 vs 1x1 strongly suggests that the binning isn't being done by the system software.  That earlier article didn't discuss pixel skipping, however.  I use "apparent" in quotes, because I'm assuming that the download is USB limited, not CPU-limited, so I'm downloading 1/4 as many bytes with 2x2.  If the software driver were doing the binning, that download would be the same data either way, so should take at least as long with 2x2, if not longer with the added processing.  If the processor or storage is the limiting resource, it might not be so clear where the binning is occurring.  I need to look at the system utilization metrics to get a better handle on this.

 

Note that I'm running on a Raspberry Pi 4B with INDI drivers.  Perhaps a bit bizarrely, the INDI configuration tab "offers" separate binning for X and Y.  I haven't explored what would happen if I tried to bin 2x3, for example.  If that "works", it would suggest that the system software is doing the bin, at least in some situations.  The image acquisition software (CCDciel) doesn't expose this option, so it will be harder to test.  I think I can do it straight from the INDI interface.

 

To the question about comparing 1x1 vs 2x2, I'm just getting started.  My first impression was a bit of a surprise, that I wasn't seeing much difference in the result other than the decrease in storage and download time.  For the test I was using StarTools Binning tool.  Pixel skipping wouldn't change the SNR in the as-downloaded image, but a subsequent bin of the 1x1 should show a 2x improvement over the pixel skipped 2x2.  I'm not immediately seeing that (at least, it's not "wow"), so that might swing the guess-pendulum back over to the software-binning-in-INDI column.  {shrug}

 

Thought to self:  Since INDI is open source, go look at the code?




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