M33 with 'Super Luminance'

[gnomus]

I have been reading (and would certainly recommend) Warren Keller's new book 'Inside PixInsight'.  In the early part of the book Mr Keller suggests that one could combine Luminance with R, G and B subs to produce a 'Super Luminance'.  I thought this was an interesting idea and tried it out with the M33 I shot last night.

This is Esprit 120 and QSI 690: Luminance - 24 x 300s; Red - 6 x 600s; Green - 6 x 600s; and Blue - 6 x 600s.

So that makes 3 hours of RGB and 2 hours of Lum.  With my Super Luminance though, have I combined 5 hours of data or only?  I also produced a version with just Lum and thought the Super Lum did look a little better.  However, I never know if this is an actual finding or if my processing just improved a tad from one version to the next.  As ever, comments and criticism are welcome.

[symesie04]

Looks a super effort, is there any chance you can post the non super luminance in here as well so we can compare?

[Barry-Wilson]

Lovely details in the core dust lanes, ruddy red HII regions in the arms are well defined.

Top drawer Steve. 

[Yamez]

Incredible shot, i finally was able to see m33 through my telescope using a 17mm Baader Hyperion last week though just a little smudge. This is super impressive, brilliant job.

[gnomus]

12 minutes ago, Yamez said:

Incredible shot, i finally was able to see m33 through my telescope using a 17mm Baader Hyperion last week though just a little smudge. This is super impressive, brilliant job.

Then you are doing better than I am - I have tried for M33 several times.  Once I thought I maybe sorta saw something, but I wasn't sure.

[vlaiv]

Great image!

And of course interesting idea of super Lum data. It does have mathematical background. If RGB filters are properly spaced and cover full L filter without much overlap (like for example Baader RGB filterset), then when adding R+G+B you are actually adding both target and sky signal (same thing would happen when doing L alone - both target and sky signal would be recorded), only down side to this is dark signal and read noise - if those are low (cooled, low read noise camera) then you gain more by having more signal then by introducing additional read/dark noise. It's a bit like software vs hardware binning - in software 2x2 binning you don't get full x2 SNR because four "doses" of read noise, so it's a bit worse then hardware binning. Same here, with L you are recording full spectrum (binned "in hardware") with single dose of read/dark noise, and when using RGB filters you are adding those together later and doing kind of software bin - each channel having a dose of read/dark noise.

So super lum does not quite have equivalent of 5h worth of L data, but a bit less than that.

[Knighty2112]

Not an imager at all, but a very nice image indeed to my eyes.  

[PatrickGilliland]

SUperlum (whichever approach you take) can be bery useful for adding that extra bit if life to an image.  You have managed just that as this look clean and not overdone.  Nice work.

Paddy

[gnomus]

Thank you very much, Paddy.

[ollypenrice]

I've played around with this somewhat. There are some extraneous issues. If you burn out your RGB stars you will struggle to find star colour, so slightly shorter RGB subs have a positive. (You don't have to apply the L to the stars if you're careful.) I've also found that, despite 'reasonable theory,' a monochrome rendered 3 hour RGB does not equal a one hour luminance. I think you have to try this for yourself to find out. I would say that with my gear I'd need at least 4 hours of RGB to match one hour of L and on the faint stuff I'm not convinced it's even as good as that. So I blow hot and cold on superluminance. At the moment I'm blowing cold. I see the processing (and maybe capture) objectives of L and RGB as being different. But, then again, Warhen is no fool!

Olly

[gnomus]

12 minutes ago, ollypenrice said:

I've played around with this somewhat. There are some extraneous issues. If you burn out your RGB stars you will struggle to find star colour, so slightly shorter RGB subs have a positive. (You don't have to apply the L to the stars if you're careful.) I've also found that, despite 'reasonable theory,' a monochrome rendered 3 hour RGB does not equal a one hour luminance. I think you have to try this for yourself to find out. I would say that with my gear I'd need at least 4 hours of RGB to match one hour of L and on the faint stuff I'm not convinced it's even as good as that. So I blow hot and cold on superluminance. At the moment I'm blowing cold. I see the processing (and maybe capture) objectives of L and RGB as being different. But, then again, Warhen is no fool!

Olly

These are good points which (as usual) I had not considered.  I think Warren also suggests throwing in your Ha as well.  I don't have the book in front of me - I'm out in the cold setting up the Ha.  I'm not sure if Ha will make much difference- I got a fair bit of red splodges just in the RGB.  If my time would be better spent, please let me know (in the next half-hour preferably!!)  

[MartinFransson]

Holy crap, that´s a nice M33!

[kirkster501]

Fabulous Steve - a great image there!

[gnomus]

37 minutes ago, kirkster501 said:

Fabulous Steve - a great image there!

Thanks Steve

[gnomus]

3 hours ago, gnomus said:

.....  I'm out in the cold setting up the Ha.  I'm not sure if Ha will make much difference- I got a fair bit of red splodges just in the RGB.  If my time would be better spent, please let me know (in the next half-hour preferably!!)  

Well the weather didn't let me down.  Clouds rolling in after 4 x 20 min subs.  Got ~15 mins into two other subs and got lost star errors, so packed in.  I have added the 80 mins of Ha and it does make a subtle difference in my opinion.  

[ollypenrice]

That's beautiful. 

I"m very reluctant to add Ha to L. Imagine doing that for the Pleiades to see why. Vrtually no Ha siganal away from the stars so you'd kill the nebulosity. Ha is at the reddest end of red which is why dropping the cyans in red in Photoshop is so effective in making it shine. On the other hand adding Ha to L in blend mode lighten might be interesting. Then it wouldn't subdue the blues. There's a test for this morning, then!

Olly

[gnomus]

10 minutes ago, ollypenrice said:

That's beautiful. 

I"m very reluctant to add Ha to L. Imagine doing that for the Pleiades to see why. Vrtually no Ha siganal away from the stars so you'd kill the nebulosity. Ha is at the reddest end of red which is why dropping the cyans in red in Photoshop is so effective in making it shine. On the other hand adding Ha to L in blend mode lighten might be interesting. Then it wouldn't subdue the blues. There's a test for this morning, then!

Olly

Hi Olly.  In the end I only got 4 x 20 min Ha subs.  I processed the 80 mins worth and then added it to the Red channel only in blend mode lighten.  This did add a few red 'flecks' to the image but, overall, I didn't think it was helping the picture - indeed it changed the general colour tone.  So I placed the non-Ha version underneath the Ha version, masked out all of the Ha and then painted in only a few Ha areas on the mask.  The change is subtle but there are a few areas that now have some red on them that were once blue. - I've highlighted 3 of the more obvious ones in the attached.    I won't add Ha to the Lum because I don't think the 80 mins stack is good enough.

 

[gnomus]

13 minutes ago, ollypenrice said:

That's beautiful.  ....

I'd missed that bit when I first read and responded to your post - thank you very much.  Your encouragement is greatly appreciated.

Steve

[ollypenrice]

In a heavily Ha dominated target using a little Ha as Lum can be helpful, but in a galaxy the Ha signal is confined to the core and to the star forming regions. I think you took the right approach and the result - little patches of accentuated star formation - is just what you'd hope for.

A throughly successful image.

Olly