Very confused by NarrowBand processing

[AlistairW]

Hello,

I am trying to get in to Narrowband, but have some questions that I think should be obvious to someone with more experience.

(I have on order a 12nm H-alpha and a 12nm OIII). (I also have a DSLR that is Baader modified and Peltier cooled).

So I think my options are to look at Bicolour initially. (Or what other options are possible - I have a CLS filter too ?).

Once I have taken the H-Alpha and OIII subs I stack as usual in DSS.

Open Photoshop and then I want to try and use this method: http://billsnyderastrophotography.com/?page_id=2358(I will need to learn all about clipping masks).

Once I have collected and stacked the H-Alpha and OIII do I need to stretch them prior to any combination, and is that all I do in PS Levels ? - Or do I use Curves as well at this point ?

Once I have gone through the workflow (i.e. combining), do I then revert to normal processing like noise reduction and sharpening etc. ?

Or if there are any good tutorials (for dummies) that I could look at, - I am sort of looking at acquisition through to processing (using PS or even Startools).

(My previous experience is simply using my DSLR a UHC-E filter and PS, so I am still unclear on channels and hubble palette - although I know bicolour is not hubble).

Thanks

Alistair 

[swag72]

When you merge bi colour images you want to assign them to the following channels

Ha - Red

OIII - Green

OIII - Blue

I rarely stretch them prior to combining them in the above channels, but have been known to if one channel is so overpowering that I just can't get a difference in colours. 

From there I use the colour select tool in Photoshop to start making the colours that I want. Very often the simple combination prior to any fiddling is quite pleasing as a more natural colour.

I then do the usual as you have said and then will always add the Ha channel (as a processed mono channel) as a luminosity layer to grab some of the detail back.

Hope that helps....... That's how I do it anyway

[ollypenrice]

Interesting that Sara combines before stretching. I didn't know that (but I only use NB to enhance LRGB.) Of course you can still play with one channel in Levels or Curves at any stage in the processing.

I stretch NB in Curves, never in levels. The only thing I stretch in Levels (by moving the grey point left) is the RGB layer. The Levels stretch is a pure log stretch which I don't find optimal for NB. For me NB responds well to a very aggressive Curve indeed. (Rather sruffily emphasized in the grab below!) The image in the background has had two of these stretches and now needs the black point moving a little to the right. My luminance curves tend to look like this as well though a little less extreme, perhaps.

Olly

[AlistairW]

Thanks for the replies - much appreciated.

I have a follow-up question. I see a lot of images from the mono CCD contingent that have HaRGB. So for me with my DSLR, I am presuming there is little point in taking individual RGB images, as the camera essentially has the colour bayer matrix. So in this instance can I image with a simple light pollution filter to give the 'combined' RGB and then add in Ha. Or am I barking up the wrong tree ?

Thanks

Alistair

[Uranium235]

Blimey Olly, thats a severe curve!

I use stock curves on every image (saved off as .acv files so I can use them whenever I want), with the occasional custom one just to tweak here and there. Why stock curves? Well, it ensures a repeatable and consistent workflow for all images.

[Pompey Monkey]

Blimey Olly, thats a severe curve!

I use stock curves on every image (saved off as .acv files so I can use them whenever I want), with the occasional custom one just to tweak here and there. Why stock curves? Well, it ensures a repeatable and consistent workflow for all images.

U235_curves.jpg

I suspect that Olly has relatively little noise to deal with!

I have used Bob Franke's simple repeated "4-11 stretch with black-point reset" and, in my very limited experience, found that it works rather well. Link here: http://bf-astro.com/non-linearcurves.htm

[ollypenrice]

My extreme curve was learned from an Adam Block video and is also used by Tom O'Donoghue, who was the first person I saw using it. I feel it has two advantages provided, as Paul says, you have the signal to support it. Firstly it really drags out the faint stuff. Secondly it preserves high contrasts. I find a log stretch of NB data gives me a rather dull result. I do have a dark sky, though, and I can imagine the extreme curve pulling up too much skyglow. It's always interesting to find that there's more than one way to skin a cat! Since Tom and I have been using these curves for years I'm surprised to find them controversial.    (I also have an entirely different 'starfield curve' which rises steeply just like the one above but then bends very low down to make a straight line all the rest of the way up. This gives you the same background sky but nice small stars.)

Edit: I don't seek repeatability. One of my axioms is that every image is different so my next step, while following a general direction, is based on what the work in progress seems to be asking for.

Thanks for the replies - much appreciated.

I have a follow-up question. I see a lot of images from the mono CCD contingent that have HaRGB. So for me with my DSLR, I am presuming there is little point in taking individual RGB images, as the camera essentially has the colour bayer matrix. So in this instance can I image with a simple light pollution filter to give the 'combined' RGB and then add in Ha. Or am I barking up the wrong tree ?

Thanks

Alistair

Yes, no point in using RGB filters when the camera already has colour filters on the pixels. Ha will only be passed by the red, so efficiency is down to 25%. It can still be done, though.

You would then be in exactly the same position as a mono imager with an RGB image awaiting enhancement from the Ha. You'll find all sorts of suggestions about how to add the Ha but mine would be to add it to red in Ps under blend mode Lighten. Ha is red and, in my book, should not be used to modify green and blue, as it will in blend mode Luminance. Sometimes, though, a hint of Ha as luminance works for me. No more than 15%.

Olly

[carastro]

I have a follow-up question. I see a lot of images from the mono CCD contingent that have HaRGB. So for me with my DSLR, I am presuming there is little point in taking individual RGB images, as the camera essentially has the colour bayer matrix. So in this instance can I image with a simple light pollution filter to give the 'combined' RGB and then add in Ha. Or am I barking up the wrong tree ?

You can use the DSLr for the colour component and then add either Ha or Luminance I have done this on a number of occasions.  The only disadvantage of this is that the DSLR is not cooled so the colour component will be more noisy, buy since the detail will be in the Ha and Luminance, you can blur the DSLr image so this gets rid of a lot of the noise (I use dust and scratches in Photoshop for this) and then use the Ha or Luminance for the detail.  However adding Ha as a luminance channel creates it's own problems as the red tends to come out salmon pink and then there are other techniques for rectifying this.  

Combining images and filters, you will need some method of making sure all the stars line up (Registering).  I use Registars for this which works excellently and this allows you to combine your DSLR and CCD images as well as they will be different sizes. 

You have to make 

Some of the methods of combining filters I find are far too complex and I found the most invaluable tutorial for combining filters was this one written by Anna Morris:

http://www.eprisephoto.com/create-lrgb

You can use this same method for combining narrowband images.  

HTH

Carole 

[MartinFransson]

However adding Ha as a luminance channel creates it's own problems as the red tends to come out salmon pink and then there are other techniques for rectifying this.

Oh, that is just what happened to me! Please, if you have a link to a tutorial on how to fix this I would be thrilled!

Sent from my phone using Tapatalk

[ollypenrice]

Oh, that is just what happened to me! Please, if you have a link to a tutorial on how to fix this I would be thrilled!

Sent from my phone using Tapatalk

It's always best to think things through from principle so you understand what you're doing.

What is luminance? It is the definitive illumination of the image. It defines what is light, what is dark, what is lighter or darker than what, once the contribution of each colour has been 'weighed' or 'balanced.'

What is Ha? Ha is that particular segment of deep red which records the hydrogen alpha emission line from excited hydrogen. The thing to remember is that it is deep red.

Now, should you illuminate the whole image as if it were all deep red? If it isn't all deep red (and it won't be) why would you do that? So don't use Ha, other than very gently, as luminance. Anything which is green or blue will be subdued by being illuminated by red light. I feel that Ha should be added to red in blend mode lighten where it will add to the red signal where it is stronger than the red. It won't affect star colour and it won't hold down genuine green and blue signal elsewhere in the image.

Olly

[MartinFransson]

Thanks Olly! I've tried different blending modes but the problem is that the deep reds turn into pink.

Sent from my phone using Tapatalk

[Pompey Monkey]

Thanks Olly! I've tried different blending modes but the problem is that the deep reds turn into pink.

Sent from my phone using Tapatalk

This page might be useful

https://starizona.com/acb/ccd/software/ps_hargb.aspx

[MartinFransson]

This page might be useful

https://starizona.com/acb/ccd/software/ps_hargb.aspx

Awesome page!! You've made my day! [emoji4]

Sent from my phone using Tapatalk

[AlistairW]

Thanks Sara , Olly , Cara for your replies - very helpful and appreciated as always !

[MartinB]

A neat trick in PixInsight is to to apply the automatic screen transfer function to each of the channels in turn. This gives an aggressive stretch but is optimised for each channel. Transfer the stretch to the histogram transformation tool and then apply this curve to the unstretched, linear channel. This gives a very well balanced combine. I like to use this as starting point but then go back and do it all over again in photoshop! I think the advantage of a single aggressive stretch, followed by a little tweaking, is that it seems to help in retaining contrast.

Plenty of ways to skin a cat!