Author Topic: Ha+OIII+Sii  (Read 11957 times)

Offline winch20010

  • Newcomer
  • Posts: 6
    • View Profile
« on: 2012 January 05 04:37:35 »
Hello Everybody,

I've just bought this software, and I'm really impressed by the 1st results I was able to produce (after a lot of hours ...)
My question is : How to combine HaOiiiSii picture?   
I actually tried the LRGB combination module, and the result is not the one I wanted.
I tried it on M42, and all the extensions are green... (I tried SNCR, histograms, ..., and I was never able to have a correct color balance)   
at the end, I removed the O3 picture, and then, the result was far better !  why ?? in M42, I had a lot of signal in O3.

So, I suppose LRGB combination is not the right way for narrowband :(   

with the o3 :

without the o3 :

thanks in advance for your help ! and happy new year !

Offline Juan Conejero

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 7018
    • View Profile
Re: Ha+OIII+Sii
« Reply #1 on: 2012 January 05 12:49:07 »
Hello and welcome to PixInsight Forum.

The toool of choice to map narrowband images to RGB channels is PixelMath. For example, suppose we have three images with the following identifiers:


and we want to combine them following the well-known Hubble palette. The PixelMath expressions would be as simple as:


respectively for the red, green and blue slots. Now suppose we want to render the narrowband data using a more 'natural' representation; for example:

SII as pure red
Ha as orange (a blend of red and green)
OIII as cyan (a blend of green and blue)

The following PixelMath expressions could be used:

0.8*SII + 0.2*Ha
0.2*Ha + 0.8*OIII

again respectively for the R/K, G and B PixelMath expressions.

In the above expressions, note that the constant factors have been defined so that no pixel can have a value larger than one (0=black and 1=white) in order to prevent saturation. For example, the first expression combines an 80% of SII and a 20% of Ha in the red channel. If the SII image is relatively weak and Ha is not saturated, as often happens, one can improve the rendition significantly by altering the expressions to remove the saturation constraint. For example, the following palette has given very good results:

SII + 0.8*Ha
0.2*Ha + OIII

You can see an example here. You'll have to vary the constants to adapt the rendition to your data, but this is the idea. Hope this helps.
Juan Conejero
PixInsight Development Team

Offline winch20010

  • Newcomer
  • Posts: 6
    • View Profile
Re: Ha+OIII+Sii
« Reply #2 on: 2012 January 05 23:01:48 »
Thank you so much for your answer Juan. 
I tried absolutely to do it with the LRGB module, and didn't think about pixelmath.
Your description is really interesting and helpful ! I was far to understand this sort of subtility with the narrowband filters ...  (photoshop didn't help me to understand...)
I made a first shoot with this technique, and the result is far better than what I tried to do few days ago :)
I'm at work now, but will try to finalize this image this evening (european time zone...)

thanks again and happy new year to everybody !

Offline kerrywaz1

  • Member
  • *
  • Posts: 68
    • View Profile
    • Spirit Bear Observatory
Re: Ha+OIII+Sii
« Reply #3 on: 2012 March 30 22:56:25 »
So whatever happened to that utility that Silvercup was writing to use sliders to adjust the Ha, OIII, SII mix?

That was a really great idea and someone needs to pick it up again and make it part of PI.
Deep Sky Instruments RC14C
FLI PL16803 w/MOAG + StarlightXpress Ultrastar
Pyxis 3"
Astro Physics AP1200
Apogee Alta F16M
Hap Griffin modified Canon EOS 5D Mark II
PixInsight, MaxIm DL, PhotoShop CS5