Author Topic: One topic that I would love to see... Narrowband and L/RGB integration  (Read 576 times)

Offline akulapanam

  • Newcomer
  • Posts: 40
    • View Profile
We need more information on the optimal way to combine narrowband with L and RGB data.  There are several methods and they are lacking good documentation.  You have your new method with the continuum map that is missing some important steps that the community has never really filled been filled in on.  A couple of us tried to fill in the gaps on Cloudynights but never got great results.  (Actually it would be nice to have a script)  Then it appears your previous method is built into the NBRGBcombination script and you also have the undocumented HaRVB-AIP and SHO-AIP scripts as well.

I have always used PixelMath to integrate HA into LUM however Warren's Inside PixInsight book suggests that you use the image integration script to do this on a noise basis instead of a factor basis.  Juan also seems to suggest that here. 

With Ha you also need to blend into G and B.  Can you generate a factor using short HB exposures?  Is there an accurate default factor (there are a couple floating around online based on scientific research). 

This would be a great tutorial worth purchasing.

Offline aworonow

  • PixInsight Addict
  • ***
  • Posts: 123
    • View Profile
I can pose a simple model appropriate for adding Ha to RGB. Assume a red filter has a width of Wr (in wavelength) and that the BACKGROUND flux in the red, through that filter, is flat across its wavelength range. Call the integrated background flux through the red filter Fr. Say that the Ha filter subtends a part of the red filter's width; denote its wavelength bandpass as Wh. By what we have already said, the integrated background flux through the Ha filter must equal Fr*(Wh/Wr). Therefore any excess above this value must be due to Ha radiation (which is shared in the observed total flux through the red filter). Consequently, we have two variables (Ha flux and background flux/unit bandwidth). The filter bandpass widths can be found from the manufacturer's specs. So, with some simple algebra, we find the oft-used, seldom explained formula that says the total Ha signal is Ha=(Wr*Fh - Wh*Fr) / (Wr - Wf), in each pixel. Having isolated the Ha signal, we can multiply it by some factor (2x, 5x, 10x) and add it to the red and make an RGB image. The image will be very red, but a BackgroundNeutralization and subsequent color calibration should restore some quite reasonable image colors.

The script NBRGBCombine utilizes this equation (although PixelMath is an easy alternative too), I believe.

Note: I would not advise using a LinearFit to restore the will undo the Ha augmentation, I think.