Is it possible to get excellent results combining LRGB with Ha without using pixelmath?

I use PI but for combining Ha I use a layers based programme such as PA, affinity or gimp. Use blend mode lighten to add the Ha to red channel and tweak the stretches of both images finely to get the result I want.

There is a script in pixinsight. 
--> Scripts – Utilities – NBRGBCombination

But you will get much better result using a pixelmath based technique. The best I know is @Charles Hagen's. 
https://www.nightphotons.com/guides/advanced-narrowband-combination/

CS

Patrice

I think this will be of interest for you. It's titled "A better way to add Ha" from the channel Sky Story. And here's his profile -> https://www.astrobin.com/users/SkyStory/

https://www.youtube.com/watch?v=CynYP_eLgRk

Noah Tingey:
I personally use the ImageBlend script (repository link: https://www.cosmicphotons.com/pi-scripts/imageblend/). It has some tweaking available with a live preview, instead of flying blind in pure PixelMath. Basic steps are:

1. Make your Ha image red using whatever method you're familiar with. (I personally use pixelmath... which I know we're trying to avoid. But this is just assigning R = Ha, so not too crazy).
2. Use curves to make the background of the Ha black, with the only bright spots being where you want the Ha to shine through
3. Use ImageBlend
1. Set your stretched LRGB image to the Base Image
2. Set your red Ha image to the Blend Image
3. Play around with the "Blend Mode" to find one that is close to what you're looking for. I can't remember which one I like but I think it might be Overlay? Or soft/hard light. Not sure.
4. Play around with the Adjustment Parameters of the Blend Image to dial it in

Editing to add: This is what my Ha image looked like after Step 2 for this image, to give you a sense for what you should be looking for:

Just to be clear.  Pure Ha signal (i.e.continuum subtracted) should not be added only to the red channel.  Remember that the two srongest spectral lines of hydrogen contribute 80% in the red (Ha-alpha) and 20% in the blue (H-beta).  That's why hydrogen emission region appear to be magenta in RGB images.

John

John Hayes:
Just to be clear.  Pure Ha signal (i.e.continuum subtracted) should not be added only to the red channel.  Remember that the two srongest spectral lines of hydrogen contribute 80% in the red (Ha-alpha) and 20% in the blue (H-beta).  That's why hydrogen emission region appear to be magenta in RGB images.

John

Interesting.   Since I already have the Ha layer in Photoshop, I can just duplicate it and assign it to blue and set the transparency to 25% of what the Red one is.   I'll give it a try.  Thanks!

Mike H - Sky View Observatory:
John, how about outlining your workflow of adding Ha to RGB. Please. I've been using pixelmath as well. 


Mike

Fortunately Vicent Peris and Martin Pugh have developed a nice way to handle this issue. The basic concept is to first subtract a scaled amount of the R channel from the Ha channel in order to remove the broadband contribution. You know it works because with the proper scaling factors, unsaturated stars in the Ha data will be removed from the data. Then a fraction of the Pure_Ha data can be added to the R channel and 20% of that can be added to the blue channel in the LRGB result. You can read a nice write up about this technique by Martin Pugh here:  http://www.arciereceleste.it/tutorial-pixinsight/cat-tutorial-eng/85-enhance-galaxy-ha-eng.  Don't forget that when you combine the Pure_Ha data with LRGB, it's necessary to multiply it by the Lum signal as well to make sure that the scaling is correct. Once you get the scaling factors plugged in, the result nicely shows the pure Ha regions.

Here's a slide from my NEIAC presentation about how to more easily remove the continuum.  Remember that you must CloneStamp out any residual stars to avoid screwing up the star field.



- John

Rick Krejci:
80%/4 blue

Hmm...  80%/4 = 20%.  You can do this in one of two ways.

1) You can add 80% of the Ha signal to the Red channel + 20% of the Ha signal to the Blue channel.

2) You can add 100% of the Ha signal to the Red channel + 25% of the Ha signal to the Blue channel.

Either approach give the correct ratio of 4:1 of the Ha signal going to the red and blue channels respectively (as I said in the slide above.)

John

John Hayes:

Rick Krejci:
80%/4 blue

Hmm...  80%/4 = 20%.  You can do this in one of two ways.

1) You can add 80% of the Ha signal to the Red channel + 20% of the Ha signal to the blue channel.

2) You can add 100% of the Ha signal to the Red channel + 25% of the Ha signal to the Blue channel.

Either approach give the correct ratio of 4:1 of the Ha signal going to the red and blue channels respectively (as I said in the slide above.)

John

Yea, the "/4" was meant to indicate 1/4 of the Red signal's 80% like I said.   Sorry, I just misinterpreted your statement,  " a fraction of the Pure_Ha data can be added to the R channel and 20% of that can be added to the blue".  It made it sound like you would only add 20% of the red channel's contribution ("that") rather than 20% the whole.      As you said, the ratio of Blue to Red is 1:4, where I got the 25% Blue/Red.    Thanks much for your insights!

Jerry Gerber:
I'm experimenting trying to combine Ha with LRGB.  I've listened to others, watched videos, read online, and everyone seems to have their own method to this madness.  

I strongly prefer to see what's happening to the image in real time, as I am working on it.  It's just my way, not right/wrong, good/bad for anyone else.  That's why I don't like fiddling with numbers if I can't see the immediate real-time results.   I guess you can say I am sense-oriented, at least when it comes to using my eyes & ears (I'm a composer). 

Enlighten me.  I'm slow...

Jerry

Seeing the effect in real time is just not a strong suit of PixInsight.  Maybe there are tricks or even magics in PI that I haven't learned (there definitely are some, since I am not a very experienced PI user), but I just don't think PI is the right choice if this is what you want.

Personally I do this in Photoshop. The layer function is perfect for adjusting and seeing the effects in real time.  You can slide the opacity bar back and forth and see the effect instantly.  You can turn layers on and off, give the layers different opacity, and compare the results with/without Ha, L or anything very quickly.

On adding Ha to LRGB, it's the best to do in the linear space, but I suppose a slightly (nonlinearly) stretched image can also work.  Basically, what I did is to use the Ha image to create a pure red image (or slightly pink, since the blue channel should contain some H-beta emission, as John mentioned above), and add it as a layer onto the LRGB image.  This of course changes the overall color balance, since the Ha image does not contain nebula Ha emission, but also all other red continuum emission from stars (including galaxies) and dust reflection etc.  So an additional adjustment layer above everything to tune down the red is necessary.

If you are familiar with Photoshop, this is really fast and easy.  And if you work on linear images, this can lead to very pleasant color.  I do this to pretty much all my HaLRGB images in recent two years.  Check my gallery to convince yourself that this really works.

Several subtle points:

1. The Ha should be added to the result of LRGB composition.  It should not be added to R before the LRGB composition.  If you add Ha to RGB first (an LHaRGB composition), the nebulas in the R channel are boosted by Ha, while the nebulas in the L channel are not.  This will lead to loss of color saturation in the nebulas.  I did LHaRGB composition before, and I learned this lesson the hard way.  Do HaLRGB, not LHaRGB.

2. When I do mobile imaging with DSLR, the above method doesn't work well.  Directly adding Ha to LRGB or RGB will boost the noise in the red, since Ha is usually not very deep from mobile imaging (always with severe time constraints).  I don't have a perfect answer on how to add noisy Ha to LRGB.  On the other hand, if you have a remote observatory whose observing time is completely controlled by you, then you can collect lot of Ha photons under moonlight, and reserve the dark periods to broadband filters.  In such cases, the Ha can be of very high S/N.  Adding such Ha to LRGB has only minor (or even no) impacts to the overall quality.  You may run an aggressive noise reduction on the Ha first before adding it to LRGB.

3. About re-adjusting color after a red Ha image is added to LRGB, in principle, the flattened 16 bit TIFF from Photoshop can be fed back to PI for the more accurate color balance and background neutralization tools. However, I don't feel this is necessary if one knows how to accurately control color in Photoshop.

Hope these help.

Or one can shoot with an Hb filter and add THAT to the B channel. Alternatively, if you are like me a purist, shoot pure RGB (in pristine skies) and you get what that nebula looks like. For OSC users there are a couple of ways, one is to use a SII-Hb filter (two pidgeons wiith one stone) the other is to use a triple band filter with a strong performance in the shorter wavelengths (e.g., L-ENH).

P.S.: Missed Wei-Hao Wang post saying exactly the same

De gustibus non disputandum est

patrice_so:
There is a script in pixinsight. 
--> Scripts – Utilities – NBRGBCombination

But you will get much better result using a pixelmath based technique. The best I know is @Charles Hagen's. 
https://www.nightphotons.com/guides/advanced-narrowband-combination/

CS

Patrice

This technique, continuum subtraction has a lot of steps but it is the easiest to control. What it does is that it uses the stacked Ha and the stacked R to create the difference, in other words what extra has Ha to offer over R.
Then you do the critical step of stretching the continuum image, so that it gets black (0.0000) everywhere except the important Ha areas. If your NB filters give you any halos around bright stars, then you can use the stamp/clone tool to remove them.
The final mix is done on the complete LRGB stretched image, without modifying its color balance. Just magically adds the Ha areas.
One disadvantage of this algorithm is that it cannot add Ha AND Oiii well. The sequence plays role, whatever you put second dominates the first. So adding Ha and then Oiii gives different results than first Oiii and second Ha. For Oiii you use either G or B do to the subtraction.
The algorithm was used here

1001 Nebulae on M33
to add Ha and Oiii.