Page 1 of 1

How does "software binning 2 x 2" DSLR RAW work?

Posted: 04 Oct 2020, 22:35
by andyho
What happens when Astroart loads a RAW (*.CR2) image using "Software binning 2 x 2"? I had assumed that the resulting RGB image sets R = R, B = B, and that G = average of the two G pixels from the RAW image. But this does not seem to be what happens.

Re: How does "software binning 2 x 2" DSLR RAW work?

Posted: 05 Oct 2020, 10:30
by fabdev
In that case the colors are still decided by DCRaw so they may be different from simple RAW R,G,B. If you need them for photometry, you really need to import the images as true RAW and demosaic inside Astroart.
Fabio.

Re: How does "software binning 2 x 2" DSLR RAW work?

Posted: 06 Oct 2020, 05:55
by andyho
Thank you, photometry is indeed the goal. I think I understand now how to debayer RAW images using the Super-pixel method with Astroart.

1. Under File > Preferences, at the RAW images tab, make sure that the option "DCRAW demosaic on load" is NOT checked. Unchecked is Astroart's default setting.
2. Under File > Open, open the RAW image as DSLR RAW;
3. Under Color > RGB demosaic, use the drop-down menu to select Algorithm "Binning 2x2" with Saturation = 0, and Hot Pixels = 0. Also, set White Balance = "None" and Bias = "None."
4. Click "OK", and the result will a debayered RGB image using the Super-pixel method;
5. Finally, select Color > Split RGB, and Astroart will generate R, G, and B monochromatic images.

The result is that the pixel values of the G image are the averages of the two G pixels in each Super-pixel (2x2 RGGB) of the original RAW. The linear dimensions of the debayered image are 1/2 the size of the original RAW, and its area is 1/4. This works perfectly for photometry, helps to save hard drive storage space, and results in faster photometric reductions.

For photometry, calibration pre-processing steps should always be done before debayering.

Re: How does "software binning 2 x 2" DSLR RAW work?

Posted: 06 Oct 2020, 11:06
by fabdev
Yes, the procedure is 100% correct.
When you need only one channel, you may also consider the Demosaic algorithms "Green channel 2x2", "Red channel 2x2", etc. which disable automatically the white balance, bias and saturation.