What's the importance of a Luminance layer?

[scotty38]

oh and holy thread resurrection batman... 🙂

[ollypenrice]

20 minutes ago, StuartT said:

If there is a need to add a luminance channel when RGB imaging, why not when SHO imaging?

As Steve says, you can use Ha or a blend as a luminance channel.

However the L RGB scenario and the NB false colour  scenario are not equivalent. The L channel catches the same information as the RGB but without the colour differentiation. That's to say, the L channel is made of red+green+blue which means it has the same components as RGB. It has no effect on the colour balance when done properly but obtains more signal in less time.

In NB imaging the three channels contain different information, which is the whole point. There is no filter (yet!) which can isolate and capture Ha, OIII and SII at the same time but without colour differentiation. If there were, it would be the NB equivalent of the Luminance filter and would save time in the same way. 

In scientific false colour imaging you would not use the strongest channel (usually Ha) for luminance because it would distort the colour balance and, therefore, corrupt the information about gas abundances in the image. (The equivalent objection applies to using Ha as luminance in broadband imaging, which is why I apply Ha to the red channel.) What you can do honestly in NB is process a copy layer of the correctly weighted and blended three channels as a luminance layer and apply it over the false colour layer.  However, the difference is that you have no more signal to play with in this synthetic lum layer than you do in the false colour layer.  In L RGB you do have more signal in L, which is why it's so effective.

Olly

[Sp@ce_d]

10 hours ago, ollypenrice said:

 

In NB imaging the three channels contain different information, which is the whole point. There is no filter (yet!) which can isolate and capture Ha, OIII and SII at the same time but without colour differentiation. If there were, it would be the NB equivalent of the Luminance filter and would save time in the same way. 

 

I haven't had time/weather/darkness to try this yet but was wondering about a dual rig with the 2600MM & MC with one of the multi band filters in each (I have tried an IDAS NBZ on the MC). Isn't that sort of (ok not quite the separate 3 channels) what Altair are claiming with thier Triband & Quadband filters? "You can also use the Quad band and Tri-Band filters as a "super luminance filter" for imaging nebulae with Mono cameras." 

 

Edit: 😲.. ok there's life in an old thread yet.. nearly 7 years old!..

Edited May 10, 2022 by Sp@ce_d
old thread resurrection realisation comment

[ollypenrice]

1 hour ago, Sp@ce_d said:

I haven't had time/weather/darkness to try this yet but was wondering about a dual rig with the 2600MM & MC with one of the multi band filters in each (I have tried an IDAS NBZ on the MC). Isn't that sort of (ok not quite the separate 3 channels) what Altair are claiming with thier Triband & Quadband filters? "You can also use the Quad band and Tri-Band filters as a "super luminance filter" for imaging nebulae with Mono cameras." 

 

Edit: 😲.. ok there's life in an old thread yet.. nearly 7 years old!..

Without having tried this I wouldn't like to say, but I can see your point. The trouble is, though, that you're not (I don't think) going to get much broadband blue data, meaning reflection components which appear in many nebulae will be blocked. I'd be more inclined to try two colour cameras, one with a multi band filter and one without. I'd then explore different ways of combining them, maybe doing a roughly 50-50 average as a first pass then adding the filtered data at full power but in blend mode lighten. Without hands on experience this remains a (moderately) educated guess. 😄

Olly

[StuartT]

Ok, full disclosure. I am still in the nursery area here. I don't use mono cameras. I'm only doing OSC imaging. But there's no harm in planning ahead right? I'll get to all this grownup black and white imaging at some point (maybe after I've mastered guiding)

Can I just ask a very basic question here? I never quite understood what 'narrowband' means. Are we talking about taking separate black and white images and then combining them in Pixisnight (or something). Or are we talking about the kind of filters I regularly use with my OSC (such as the Optolong L Extreme) which restrict the image to 'narrow bands' (i.e. hydrogen alpha and O iii)?

 

[carastro]

Narrowband are wavelengths from a different part of the light spectrum. The sky is rich in these wave lengths such as Hydrogen alpha Oxygen3 and Sulphur 2.  A OSC camera is not very good at picking these up as it is designed for broadband colours ie the colours seen with the naked eye and known as RGB, but a mono camera with the relevant filters does a good job of capturing narrowband. 
These days there are fancy filters which can help OSC cameras to pick up these wavelengths.  

Narrowband are combined in post processing to replace Red, Green and Blue  making a coloured image   This can be done in differing ways such as HOS, HSO, SHO and HOO In place of RGB.  (H = Hydrogen, O = Oxygen and S = Sulphur, RGB = Red, Green and Blue). and the resulting image is known as a false colour image   NB: SHO is also called Hubble Palette. 
 

H is usually referred to as Ha, O is usually referred to as Oiii and S is usually referred to as Sii. 
 

Narrowband wave lengths are good for excluding light pollution which is in a different part of the light spectrum.  

Hope this helps to explain a bit. 
 

Carole

Edited May 12, 2022 by carastro

[DaveS]

I suppose you *could* use one of These as a NB Luminance filter, but would you *really* want to pay 4 figures?

[Sp@ce_d]

2 hours ago, DaveS said:

I suppose you *could* use one of These as a NB Luminance filter, but would you *really* want to pay 4 figures?

Interesting.. but its not that much more than a single band 2" 3nm Chroma or Astrodon! At 4nm it sits in-between most options of 5nm or 3nm for a single filter. Bearing in mind a 2" 5nm chroma OIII sits at £900 or £1200 for 3nm, it starts to look cheap compared to a set of single filters!

[StuartT]

On 11/05/2022 at 09:17, carastro said:

Narrowband are wavelengths from a different part of the light spectrum. The sky is rich in these wave lengths such as Hydrogen alpha Oxygen3 and Sulphur 2.  A OSC camera is not very good at picking these up as it is designed for broadband colours ie the colours seen with the naked eye and known as RGB, but a mono camera with the relevant filters does a good job of capturing narrowband. 
These days there are fancy filters which can help OSC cameras to pick up these wavelengths.  

Narrowband are combined in post processing to replace Red, Green and Blue  making a coloured image   This can be done in differing ways such as HOS, HSO, SHO and HOO In place of RGB.  (H = Hydrogen, O = Oxygen and S = Sulphur, RGB = Red, Green and Blue). and the resulting inage is known as a false colour image   NB: SHO is also called Hubble Palette. 
 

H is usually referred to as Ha, O is usually referred to as Oiii and S is usually referred to as Sii. 
 

Narrowband wave lengths are good for excluding light pollution which is in a different part of the light spectrum.  

Hope this helps to explain a bit. 
 

Carole

Thanks Carole. So would it be correct to say that an OSC and Optolong L Extreme filter would be 'narrowband'? Or is the term reserved only for mono cameras shooting the separate filters?

[vlaiv]

On 10/05/2022 at 09:03, StuartT said:

If there is a need to add a luminance channel when RGB imaging, why not when SHO imaging?

It really depends what you are trying to achieve.

It is important to understand why is luminance useful in the first place. Here is demonstration of that.

Let's take nice image and see what luminance and color information really represent - for example this image:

When we decompose image into components - luminance only and chrominance only:

You can see how much information is carried by luminance and how much by chrominance components. Lum carries much more information for our visual system.

This is very important for our perception of noise.

Look at these two images for example - first one is our baseline image with noise added to luminance channel, and second is base line image with same amount of noise added to chrominance channel (we could even argue that we added more noise in second case as we applied noise to two channels of chrominance - not only one):

Difference is obvious.

It is therefore beneficial to have higher SNR to luminance and spend more time to capture luminance information that chrominance component (out of total imaging budget). It is also important that we process data in proper way for us to actually benefit - we really need to treat luminance as luminance information (in appropriate color space)

What about NB imaging then?

There are two approaches here.

First - we must understand that by doing nothing and just using pure SHO palette - we are already doing a lot to mimic L benefit. How so?

It turns out that luminance is not composed out of RGB components equally  - red, green and blue as defined in sRGB standard and used in our display devices don't carry same amount of luminance.

If we take pure sRGB primaries - like in image above and we convert that to grayscale image (just luminance) - we get this:

Green carries the most luminance information, followed by red and blue in the end (red and blue are close - but dimmer than green).

Ha signal is often the strongest out of usually imaged triplet - and thus has best SNR if we dedicate equal amounts of time. In normal SHO palette (not modified like we often see in images for aesthetic effect where people "kill off" green for some reason) - Ha maps to component that carries the most luminance information. This is good thing as it makes image look less noisy visually - just by choice of mapping. This of course works only if one does not kill off green or change palette substantially from original SHO (most of what people say is SHO is not SHO at all).

Back to luminance for NB imaging and two approaches:

1. approach is to use Ha as luminance information. This works well because again - Ha is the strongest of the three and usually is present all over the place. One just must take care if there are areas with strong OIII that does not contain much Ha signal - as in that case OIII will be dimmer than it should

2. approach is to use quad band NB filter. This works with mono cameras basically the same as LRGB for broad band imaging. One should spend most time with quad band filter and then take small amount of separate NB data just for "coloring" of the image (same as with LRGB really).

In either case (any time we use luminance) - we need to make sure we apply it properly to the image - otherwise we are not going to get optimal results.

[AstroMuni]

1 hour ago, vlaiv said:

It is important to understand why is luminance useful in the first place. Here is demonstration of that.

Great explanation as usual vlaiv 🙂

[StuartT]

1 hour ago, vlaiv said:

It really depends what you are trying to achieve.

It is important to understand why is luminance useful in the first place. Here is demonstration of that.

 

Thanks so much for this @vlaiv - I always learn so much from your answers! 👍

[carastro]

Quote

OSC and Optolong L Extreme filter would be 'narrowband'? 

As I have used Mono cameras since 2012 and no experience of OSC and these filters that will enable narrowband wavelengths to be captured, I can't really give a confident answer on that,  but I guess so.  I could be wrong but my understanding is that these type of filters only capture Ha and Oiii.  They give pretty good results from what I have seen.

Carole 

[Sp@ce_d]

6 hours ago, StuartT said:

Thanks Carole. So would it be correct to say that an OSC and Optolong L Extreme filter would be 'narrowband'? Or is the term reserved only for mono cameras shooting the separate filters?

 

2 hours ago, carastro said:

As I have used Mono cameras since 2012 and no experience of OSC and these filters that will enable narrowband wavelengths to be captured, I can't really give a confident answer on that,  but I guess so.  I could be wrong but my understanding is that these type of filters only capture Ha and Oiii.  They give pretty good results from what I have seen.

Carole 

and of course there's nothing to stop us using any filters in front of a OSC.. you could have the same filter wheel full of filters as a mono. All a filter does is allow a wanted range of emissions through & blocks the rest. A narrower "notch" in the emission "window" tends to be called Narrowband. The difference using them with OSC is that it has a CFA (colour filter array.. most common type is the bayer matrix) over the sensor. So using an Ha filter as an example would only pass through to the red pixels on a OSC. I guess using a OSC in that way would become a 3SC for SHO 🤣. So back to using a multi band filter, I've used the IDAS NBZ with the 2600MC in as yet limited testing & I've been quite impressed. Depends on the target of course. Its a bit different to process as with true mono with individual filters from my experience so far.. & with mainly filters that pass Ha & OIII its a kind of HOO rather than SHO. As I understand it from the things I've read through & managed to digest.. (I'm sure @vlaiv has this stuff for breakfast )  the CFA on OSC will reduce some of the SNR generally but with narrowband there is more loss of high frequency information. 

[StuartT]

3 hours ago, carastro said:

As I have used Mono cameras since 2012 and no experience of OSC and these filters that will enable narrowband wavelengths to be captured, I can't really give a confident answer on that,  but I guess so.  I could be wrong but my understanding is that these type of filters only capture Ha and Oiii.  They give pretty good results from what I have seen.

Carole 

I can definitely vouch for the L Extreme!

https://www.astrobin.com/wm7bzk/

https://www.astrobin.com/z8ufiy/B/

https://www.astrobin.com/phln6z/