Entering the Field of HaRGB - IC405 as a test. Does my approach make sense?

[graemlourens]

Hi everybody.

As i'm little fed up with tricolor SHO narrowband i'm switching to something i have never done yet. HaRGB. I've picked IC405 as the perfect target as its got strong HA and a twitch to it that there is a reflective blue part. Its a beautiful target and now i'll have to learn a few basics. Maybe some of you can help me.

My approach is to

- get a good HA master (done)

- get a decent set of RGB without blowing out the stars, using 120 seconds (done). This data will pose as a 'baseline' to enhance the red (& blue)

- get a decent set of longer Blue exposures (10 minute frames). I'll use this data to blend it into the previous blue master.

I don't want to blow out my stars, and the Atik460 doesn't have a high full well capacity, so stars are out very fast (at my focal length).

Is this the right approach? I'll be probably using Photoshop for most of this work, as it seems easier to handle than PI.

 

This is a testpicture, quickly autostretched HA and put it into the RGB masters (not got the longer blue exposures yet). 

 

Any other hints i could use while blindly stepping into this new field?

Kind regards, Graem

[ollypenrice]

Thanks for the PM.

I think that messing about with the RGB balance will be risky. I always shoot equal amounts of R,G and B. Often there are broad spectrum, natural coloured features all around an image, not just in the stars. You mention the blue reflection nebula in the Flaming Star but most images contain something blue or dusty brown besides the stars. If you shoot extra blue without extra green you'll get some strange blues, I think.

I processed some Atik 490 data once with a guest and was struck by the lack of star colour, which I put down to the shallow well depth. I can shoot 10 minute RGB subs and find, with my Atik 11000, that my stellar cores are mostly below 65000 counts even once stacked. Of course I'm not filling the chip with LP here either. I'm not sure how important this is since you can pull outer star colour into the cores, the easiest way being by using Noel Carboni's Actions. (Astronomy Tools by ProDigital should Google happily.)

My own HaLRGB images set out not to be 'red Ha images with full star colour' but to be full-on boradband images enhanced by Ha. So I aim for a good RGB to start with. What about Luminance? I had a spell of not using any on emission-dominated subjects but have concluded (as Tom argued all along) that you do benefit from L to lift the RGB before the Ha goes in.

I'll copy my own Ha combination method from another thread:

Get the colour image finished and the Ha image finished. Co-register them to fit each other. You can, if you like, push the Ha a tad harder than you would for a standalone presentation because most of the noise will be excluded from the final image by this method:

Split the colour channels and paste the Ha over the red. Change the blend mode to Lighten. Flatten and recombine the colours.

Your Ha has now been applied to the red only where it was lighter than the red, leaving the stars entirely unaffected, all being well. Save this as HaRGB.

If you paste this over the original RGB you can see the difference and remove any artefacts you don't like using the eraser.

You can try a little Ha as luminance as well but I would never exceed 15% or so because you will light the full colour spectrum in red light, which just can't be right. In general Ha as luminance in DS imaging should be a last resort in my view and done minimally. Adam Block describes a similar method but adding Ha to red in Blend Mode Screen and then black clipping it. I have a video but haven't yet tried this.

If, when your Ha is pasted over your red in blend mode lighten, it doesn't seem to be doing much, you can give it a lift in Curves until you see it coming into play. Hold down the low Ha to stop it turning the background red when you do this. (You will have stretched your Ha to the noise limit before combining it, but the noise limit is defined by the low Ha brightnesses which will almost certainly be below the level of the reds, so the noisy parts won't be applied in blend mode lighten. Cute!) You can also process your Ha for exaggerated contrast because the contrasts will be diluted when blended with red. ANother trick with Ha destined for red is to do a very aggressive star reduction on it. You don't want the Ha to brighten the stellar reds at all.

Some people advocate adding the Ha to red a little at a time. I can see no point in this. You might as well whack it on at full power because, to dial it back, all you do is paste the HaRGB onto the RGB and then dial back the opacity till you get the Ha contribution you like.

Olly

 

[kirkster501]

I shall investigate further.  Seems there are many different approaches to this.  What I was going to do was blend the HA with the Red and then use that combined Ha/Red as the "red" when I do RGB.  Then use Luminance as per nornal.  Another methid I read was to blend Ha with Lum and use that as the "Lum" to then combine with RGB...

 

[ollypenrice]

1 hour ago, kirkster501 said:

I shall investigate further.  Seems there are many different approaches to this.  What I was going to do was blend the HA with the Red and then use that combined Ha/Red as the "red" when I do RGB.  Then use Luminance as per nornal.  Another methid I read was to blend Ha with Lum and use that as the "Lum" to then combine with RGB...

 

The second method went out of fashion about eight to ten years ago when everyone got fed up with pink images gleaming with bloated blue stars!!!

Ha stars are smaller than red, so if your blending of Ha with red reduces the star size in red you're going to get funny stars, lacking red round the edges in the colour mix and causing the blue to dominate. That's why blending Ha to red in 'Lighten' mode is so attractive. The Ha won't lighten the red for the stars because the Ha stars are smaller.

But we don't all want to be doing the same thing or there will be no progress. Experiment is everything.

Olly

[graemlourens]

20 hours ago, ollypenrice said:

Thanks for the PM.

I think that messing about with the RGB balance will be risky. I always shoot equal amounts of R,G and B. Often there are broad spectrum, natural coloured features all around an image, not just in the stars. You mention the blue reflection nebula in the Flaming Star but most images contain something blue or dusty brown besides the stars. If you shoot extra blue without extra green you'll get some strange blues, I think.

I processed some Atik 490 data once with a guest and was struck by the lack of star colour, which I put down to the shallow well depth. I can shoot 10 minute RGB subs and find, with my Atik 11000, that my stellar cores are mostly below 65000 counts even once stacked. Of course I'm not filling the chip with LP here either. I'm not sure how important this is since you can pull outer star colour into the cores, the easiest way being by using Noel Carboni's Actions. (Astronomy Tools by ProDigital should Google happily.)

My own HaLRGB images set out not to be 'red Ha images with full star colour' but to be full-on boradband images enhanced by Ha. So I aim for a good RGB to start with. What about Luminance? I had a spell of not using any on emission-dominated subjects but have concluded (as Tom argued all along) that you do benefit from L to lift the RGB before the Ha goes in.

I'll copy my own Ha combination method from another thread:

Get the colour image finished and the Ha image finished. Co-register them to fit each other. You can, if you like, push the Ha a tad harder than you would for a standalone presentation because most of the noise will be excluded from the final image by this method:

Split the colour channels and paste the Ha over the red. Change the blend mode to Lighten. Flatten and recombine the colours.

Your Ha has now been applied to the red only where it was lighter than the red, leaving the stars entirely unaffected, all being well. Save this as HaRGB.

If you paste this over the original RGB you can see the difference and remove any artefacts you don't like using the eraser.

You can try a little Ha as luminance as well but I would never exceed 15% or so because you will light the full colour spectrum in red light, which just can't be right. In general Ha as luminance in DS imaging should be a last resort in my view and done minimally. Adam Block describes a similar method but adding Ha to red in Blend Mode Screen and then black clipping it. I have a video but haven't yet tried this.

If, when your Ha is pasted over your red in blend mode lighten, it doesn't seem to be doing much, you can give it a lift in Curves until you see it coming into play. Hold down the low Ha to stop it turning the background red when you do this. (You will have stretched your Ha to the noise limit before combining it, but the noise limit is defined by the low Ha brightnesses which will almost certainly be below the level of the reds, so the noisy parts won't be applied in blend mode lighten. Cute!) You can also process your Ha for exaggerated contrast because the contrasts will be diluted when blended with red. ANother trick with Ha destined for red is to do a very aggressive star reduction on it. You don't want the Ha to brighten the stellar reds at all.

Some people advocate adding the Ha to red a little at a time. I can see no point in this. You might as well whack it on at full power because, to dial it back, all you do is paste the HaRGB onto the RGB and then dial back the opacity till you get the Ha contribution you like.

Olly

 

Olly thank you very much for your extensive reply. I do have to study a few details of it before i can come up with more detailed followup questions. 

The testpicture i made in the post above as indeed done by adding HA to the red channel by 'Lighten' Mode, so i guess i did this intuitively the correct way (pat on my back). It seems very 'simple' though, and the time it took me to process was very short compared to what i'm used to with narrowband tonemapping. It feels like i'm skipping 90% of the steps, but i guess i'll find those out in time.

My thought of taking longer Blue data is just to blend in that certain part of the reflection of the blue star. I get your point in not messing with the balance, but do i understand correctly then that with my shallow well depth i will be always limited to rather unspectacular RGB images? Or will i rather have to in some way shoot longer RGB (where the stars are blown out) and then shoot lower exposure shots, and somehow just add those stars to the longer exposure?. Not quite sure what path to take here.

To experiment further, i have now taken those longer Blue shots, as well as another longer set of Luminance (5 minutes, whereby the stars are blown out a lot). Now i have all the data for all the different paths, and will experiment.

One question that pops to my mind: Is it 'done' that if you have slighly blown out stars (or very highly exposed, so that color doesn't really show) that you reduce the brightness of the stars, and shrink them, afterwards like you said 'pulling' the color back into the core. Is this standard practise in RGB images or considered too much tampering?

Kind regards, Graem

[graemlourens]

I've been further testing and have a million questions. Somehow have to logically sort them before i ask.

I'm not happy though yet with the result. Is this anything worth looking at? I can't tell as i've not really seen many HaRGB's yet, as i was concentrating on narrowband only.

I wasn't getting that blue reflective light, as my subs are just very short (120s). I did a set of 600s blues just to compare and it was much better, but obviously stars totally blown out. I didn't do a set of G & R 600s yet, but will still do that to see if there is any way of keeping that star color. In the meanwhile i, with few masks & trickery, just took a little of the blue reflective part and enhanced the Blue channel of the final RGB image just to pop it out a little more, but in a real target i would not spend nearly 2 ours just to pop that small part a little further. I'm interested to see how this will compare to a full 600s RGB.

Maybe i would have to focus on shooting shorter star subs for color, and longer rgb subs for the nebulosity (HA is not a problem), any hints on if that is 'standard practise'  (and maybe some tips or links?) i have never successfully replaced stars in a longer sub with ones from shorter subs. I guess with masks its possible, but not trivial.

Any insight welcome. Not really found my way here yet

Kind regards, Graem

 

[ollypenrice]

I never have shot short subs for star colour. I dare say you could and that you could, with care, blend them. But I wonder if you need to? Noel's Actions can pull in edge-of-star colour to the core, which might be all you need.

I can't speak for anyone else but it's certainly my standard practice a) to hold down stars in stretching and b ) further reduce them in later stages of processing. This means reducing both brightness and extension, so making them more colourful.

I think your first test is going well. I would push the Ha harder though. Stretch it more and boost the contrasts. If it is just going into the red channel you can be a real thug with it! You really would not believe what the Ha looked like for this image when I blended it with red. It was an abomination! I had to stretch the lower end right up and white clip the brightest half of the data so it looked like nothing on earth, but I just kept in mind what was, and what was not, going to end up in the final red channel. I was just dead set on doing a Double Cluster with Ha background, come what may.

Olly

[graemlourens]

13 hours ago, ollypenrice said:

I never have shot short subs for star colour. I dare say you could and that you could, with care, blend them. But I wonder if you need to? Noel's Actions can pull in edge-of-star colour to the core, which might be all you need.

I can't speak for anyone else but it's certainly my standard practice a) to hold down stars in stretching and b ) further reduce them in later stages of processing. This means reducing both brightness and extension, so making them more colourful.

I think your first test is going well. I would push the Ha harder though. Stretch it more and boost the contrasts. If it is just going into the red channel you can be a real thug with it! You really would not believe what the Ha looked like for this image when I blended it with red. It was an abomination! I had to stretch the lower end right up and white clip the brightest half of the data so it looked like nothing on earth, but I just kept in mind what was, and what was not, going to end up in the final red channel. I was just dead set on doing a Double Cluster with Ha background, come what may.

Olly

Thank you olly for your feedback.

I did try Noel's Actions, and the Star color action does work well indeed. A few questions i have though:

Assuming you're doing Ha(L)RGB - when you're doing starcontrol, pulling color back into the stars etc, are you doing all this after combining L with the RGB? As i've always had troubles when only taking care of L because you'll have to match the star size in RGB or you'll have colorful haloes. I've come to see that doing this twice will never really match up. But maybe its really only done after combining Lum?

But that would mean that you only stretch a little, combine Lum and then continue stretching, and this is the opposite of what i'm used to with a Luminance layer.
For me its all boiling down to this order of events, mostly concerning the stars.

Would you be able to give a quick step order assuming Ha(L)RGB assuming one has to 'handle' the stars extensively (color, size etc)?

Much appreciated... Kind regards, Graem

[ollypenrice]

3 hours ago, graemlourens said:

 

Assuming you're doing Ha(L)RGB - when you're doing starcontrol, pulling color back into the stars etc, are you doing all this after combining L with the RGB? As i've always had troubles when only taking care of L because you'll have to match the star size in RGB or you'll have colorful haloes. I've come to see that doing this twice will never really match up. But maybe its really only done after combining Lum?

 

Much appreciated... Kind regards, Graem

I just don't find this. Small stars in L will hold down larger RGB stars quite happily for me without creating haloes. I might reduce the RGB stars before adding L but only after running the star colour action. That needs to be done first on the 'natural' stars. I will star reduce the L layer and then maybe further reduce the LRGB stars when combined.

I give my RGB a very simple log stretch, moving the mid point slider to the left.  I do more 'custom' curve stretches in L, often blending different stretches in Layers. I avoid pushing the L stretch to more than about 80% of what it will give, though, or it overwhelms the colour. I do the final hard stretch to the LRGB.

Olly

[graemlourens]

On 1/3/2017 at 09:56, ollypenrice said:

I just don't find this. Small stars in L will hold down larger RGB stars quite happily for me without creating haloes. I might reduce the RGB stars before adding L but only after running the star colour action. That needs to be done first on the 'natural' stars. I will star reduce the L layer and then maybe further reduce the LRGB stars when combined.

I give my RGB a very simple log stretch, moving the mid point slider to the left.  I do more 'custom' curve stretches in L, often blending different stretches in Layers. I avoid pushing the L stretch to more than about 80% of what it will give, though, or it overwhelms the colour. I do the final hard stretch to the LRGB.

Olly

Interesting, so i must be doing something wrong OR i am trying to reduce the star size too much (with PI i am able to reduce them by quite a lot without artefacts, and i mostly do this, maybe this is the cause of why i have problems. The stars in my ED80 are fairly large).

I will seriously have to sit down and re-think a few steps, and evaluate how and what i'm processing. Its really like starting from scratch, as its so massively different from narrowband, especially tonemapping.... Will come back to you with specific questions as soon as i can put them together, but now i've got some homework to do first...

Kind regards, Graem

[graemlourens]

I decided to simplify a few things first so i can concentrate on basics. This lead me to forget about shooting Luminance separately. As i have to handle my stars very intensively (probably because of Seeing conditions & the ED80) this is an extra complexity layer and i'll maybe come to this at a later stage.

I have experimented a lot, and with the current data i have (the RGB only having 120s exposures is very limited) and came up with this, that i can only half accept. I don't like the fully saturated red, so i decided to combine the extracted Lum with HA just slightly, to get it away from the pure red madness.

Second thing i have found out, is that i will not get very far with the Blues if i do not shoot the RGB data with much longer exposures (maybe 600s). I will now re-shoot G & R with 600 sec exposures (i already have a set of blues) and will the invest time in fixing the mostly overexposed stars. Reading around there are many ways in retreiving back the colour (additionally will be using the Photoshop plugin Olly referred to earlier on). 

If i stay on short exposure RGB data, and enhancing it with Ha, it will always be a pure red image with few small & colourful stars, and i do not want to accept that.

Will post my progress when i've gotten the longer exposed Green & Red data, and try this all over again!

Kind regards, Graem

[ollypenrice]

2 hours ago, graemlourens said:

If i stay on short exposure RGB data, and enhancing it with Ha, it will always be a pure red image with few small & colourful stars, and i do not want to accept that.

Exactly right.

Olly

[graemlourens]

On 1/8/2017 at 17:23, ollypenrice said:

Exactly right.

Olly

Ok so finally skies cleared up just enough yesterday for me to gather the longer 600s subs for Green & Red (blue i already had)

Using these subs, as well as fixing the overexposed stars (lots of them) and using Noels Increase star colour action, i must say that i have lost the fear of only shooting low exposure RGB frames, as it is possible to keep the star colour indeed.

Obviously crisp stars is even more difficult with a Scope like the ED80, but that will have to be something i accept and fix as good as possible. Still it is very hard to get a nice defined blue reflection, as if i overlay HA too strongly (as that area is also very strong in HA) it just turns very magenta.

This is my effort, and i think it is much more pleasing to the eye than the previous efforts (that were pretty much just red...)

The blue reflection part though did not automatically 'pop up' as i expected, but as i had much better SNR i was able to saturate the blues in that area slightly more to get the contrast i was looking for.

What do you think?

[ollypenrice]

That looks good to me. The blue reflection is always difficult but now you have it.

Olly