Processing Veil Bicolor in Pixinsight

ArmyOrtho
04 Jun, 2018 22:00
Hello everyone!

I've found myself stuck with processing yet again and wanted to ask cor a consensus view on some steps to take to get the result I'm looking for.

I've collected about 2 hours each of the Veil Nebula in widefield using Ha, OIII, and SII filters from my suburban site.  I was pleased with the individual collections, but when processing it, I can't seem to get the typical "veil bicolor" look that I was expecting.  I had it once when screwing around with color combination, but haven't been able to find it again.

The image I posted earlier  (https://astrob.in/349896/0/) consists of an LGRB combination of Ha on the red channel, OIII on the blue channel, and (OIII*.9)+(Ha*.1) on the green channel.  (I listed it as an HSO image, only to realize after I posted that there is no SII in it)  This math was given to me for a different shot I attempted in narrowband by a very helpful redditor who told me it was where he started to get a "more natural" color to a bicolor shot.  To this, I agree, but I'm a stuck when tring to make the HOO combination come back with the red and blue  typically seen.

Can you guys point me in the right direction?

Many thanks!
RickS
05 Jun, 2018 03:57
I often start with a straight HOO combination (use LinearFit to equalize Ha and Oiii first or apply an unlinked stretch to balance them.)  CurvesTransformation with a Hue curve will let you push the cyan towards blue.  The ColorMask script and a few more curves will let you tweak the colours to taste if more is needed.

Here's an example done this way: https://www.astrobin.com/340790/?nc=user

This is the Veil in bi-colour using a somewhat more sophisticated technique: https://www.astrobin.com/143114/?image_list_page=3&nc=&nce=  For this one I used the value of the signal in Ha and Oiii to vary the hue using a bit of PixelMath.

Cheers,
Rick.
ArmyOrtho
05 Jun, 2018 04:05
Many thanks.  I was screwing with pixel math when this all started, but didn't use linear fit.  I'm sure that's what my issue was.

Do you remember the math you used to get that second image?  Both are absolutely amazing!
RickS
05 Jun, 2018 09:41
William Jordan
Many thanks.  I was screwing with pixel math when this all started, but didn't use linear fit.  I'm sure that's what my issue was.Do you remember the math you used to get that second image?  Both are absolutely amazing!

LinearFit is a great way to match the NB channels so that you have a balanced starting point and some predictability to the results.

Thanks for the kind comment.  For the Veil I used multiple PixelMath formulae to build CIE L, c and h (lightness, chrominance and hue) images for the Ha and Oiii and used ChannelCombination to combine them into colour images which I then blended again with PixelMath.  I have some notes but exactly the same formulae probably won't work with different data.  I have been intending to write a script that allows simple experimentation with this technique and had some recent thoughts on how to do it, so this is timely.  Will see if I can find a few spare hours to work on it.

Cheers,
Rick.
2ghouls
05 Jun, 2018 17:56
William Jordan
I was screwing with pixel math when this all started, but didn't use linear fit.

LinearFit will definitely help. You can also try: combining with PixelMath (while linear) then stretching VS. stretching the channels then combining with PixelMath. I find the latter usually works better for me, but I have mostly done tonemapping which requires stretching first. For the Veil, you probably don't need to tonemap as the OIII signal is quite strong. That said, I feel you might have more control over how much the blue shows up by stretching (and using Curves) then combining because you can better control the result that way.

If the above wasn't clear, here are the steps I am suggesting:
1. Run NoiseEvaluation and LinearFit to the channel with the least noise (Ha to OIII for instance)
2. HistrogramTranformation using small steps (do not use STF AutoStretch) then Curves Transformation. Just assess visually until you get the OIII nebulosity as bright or brighter then the HA so it stands out.
3. PixelMath to combine. You can use the formula you mentioned or Veil could probably just be straight HOO.

Nico
 
Register or login to create to post a reply.