super Luminance on all RGB SHO channels?

I always use a SuperLuminance when doing LRGB. The gain in SNR is measurable. Unless your RGB subs have higher FWHM’s I can’t think of a reason not to do a SuperLum.

With Narrowband I usually do a PseudoLuminance. Unlike the SuperLum where you have ImageIntegration do a weighted combination of the LRGB masters, with Narrowband I combine the channels manually in Pixelmath using factors. The idea here is that we generally reduce the strength of the Ha so much trying to get NB colors balanced that the SNR is seriously degraded. So I usually do something like Ha*.5 + OIII .25+ SII*.25 to try and lean more on the Ha for SNR. But you have to be careful that there is enough of the O and S in there to support their particular structures and not have colors with no Luminance behind them. Even equal amounts of S, H, and O will have a better SNR than your SHO color mix though.

If I’m doing SHO+RGB I’d probably test some combinations of extracted L from the RGB master and the SHO masters.

Thanks for your advices and sharing your experiences, guys.

Apart from Kevins advice to always use synthetic Luminance for narrowband images - are there alternatives to benefit from the SNR of a superluminance?

How do you deal with the issue that the Lum channel or any super Luminance are less sharp than the SHO image, yet have much better SNR.

BlurXTerminator I tried but did not make substantial change.

So what is best practice for making use of a Lum channel with SHO images?

Arny

Arny · Sep 15, 2025, 08:47 AM

How do you deal with the issue that the Lum channel or any super Luminance are less sharp than the SHO image, yet have much better SNR.

Arny

I want to make sure I understand your question. Are talking about adding Luminance data captured with a Luminance filter to SHO color data?

If so I would not add pure L data as a Luminance channel when the color data is pure SHO. Differences in sharpness would not really be the issue there but the differences in the actual structure of the object matching up. If my color data is pure SHO, I would create a PseudoLuminance from the SHO masters but not use actual Luminance data.

If I have RGB and SHO data in the color, and I want to use L data too, I would create my RGB+SHO color mix first. Then I would create a SuperLuminance and subsequently add the SHO data to the SuperLum in pixelmath using some multipliers depending on the relative strengths of the channels.

For example, RGB/K: SuperLum*.5 + H*.2 + O*.15 + S*.15

I would then use that result as the Luminance channel for my RGBSHO color data.

You mentioned a difference in sharpness. Why would there be a difference in sharpness between the L and SHO data? Unless the seeing was quite different on the nights you shot your L and SHO they should have pretty close FWHM’s.

Kevin

OK, maybe I understand better what you are after. If you want to get all the benefits of the higher SNR of Superluminance but retain all of the narrowband structures you need to add continuum subtracted narrowband masters to your SuperLuminance. You do that in linear space with starless, continuum subtracted Narrowband masters.

I outlined almost that exact set of steps in my last image here: https://app.astrobin.com/u/morefield?i=a9abdt#gallery

The only difference is that I didn’t use Luminance filter data. So in step 8 where I used extracted L from the RGB master, you would instead use ImageIntegration to create a SuperLuminance from the LRGB masters.

Does that sound like what you are looking for?