Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Ratio for Luminance to Color data when shooting Narrowband

  • Please log in to reply
69 replies to this topic

#1 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 09:19 PM

Hey all, 

 

I have found that luminance in HA, or a stacked combination of all my data (oiii, ha, sii) results in really really good clean final images with almost zero noise and great color. 

 

One question I am finding myself asking is how much of a ratio should we be aiming for? Should we do 1:1 lum to color? (lum being ha or triad) or should we aim for 2:1 lum to color to get better, more crisp detail where the color is filled out after (x) hours sufficiently? 

 

I just shot this pacman nebula  and did roughly 11h in Ha, and 4h in Oiii. At this point to improve more, what would be the data to gather? I don't really plan on doing Sii as I like the color palette with this HOO rendition. 

 

Should I be adding more Ha for Luminance or more Oiii? 

 

What ratio in general should you aim for with Lum to Color data (oiii/sii)? 

 

And finally has anyone used the TRIAD ultra for luminance channel? Do you find it better/worse/no different than a stack of all narrowband data? 

 

Thanks!



#2 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,718
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 30 November 2020 - 09:24 PM

Don't use luminance with narrow band. It basically defeats the purpose. The entire goal with NB is the extreme contrast between channels. If you combine a luminance channel, you lose most of that contrast, rendering the efforts to acquire all that clean, high contrast data largely moot. This goes for synthetic luminance as well. 

 

You should combine each channel on their own, possibly with some slight custom blending across channels (i.e. PixelMath in PI) and leave it at that. 

 

If you want to improve what you've got, then more data is needed. Keep in mind, you need 4x as much data as you have, to get 2x the SNR... So if you have 11 hours now, you need around 44 hours to really get a reasonable improvement...you'll have to judge whether that is worth it or not for you. That said...there is a TON of additional faint signal around Pacman there that very, very few people pick up. I one-handed count at most. 


  • rayp likes this

#3 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 09:29 PM

Don't use luminance with narrow band. It basically defeats the purpose. The entire goal with NB is the extreme contrast between channels. If you combine a luminance channel, you lose most of that contrast, rendering the efforts to acquire all that clean, high contrast data largely moot. This goes for synthetic luminance as well. 

 

You should combine each channel on their own, possibly with some slight custom blending across channels (i.e. PixelMath in PI) and leave it at that. 

 

If you want to improve what you've got, then more data is needed. Keep in mind, you need 4x as much data as you have, to get 2x the SNR... So if you have 11 hours now, you need around 44 hours to really get a reasonable improvement...you'll have to judge whether that is worth it or not for you. That said...there is a TON of additional faint signal around Pacman there that very, very few people pick up. I one-handed count at most. 

Yeah I'm just not sure I agree. I have processed both with and without a luminance channel added in, and when adding a luminance channel I consistently get cleaner results. Additionally most of the bigger astro people I follow are doing this as well (like ak_astro just as an example who thoroughly documents their PI flows and get insane results) with narrowband imaging mono imaging. I get excellent contrast with an added luminance channel, the only real difference I see between my lum and non lum processing is more noise in the final image. 

 

All this being said it's totally possible there is some PixInsight magic I am not aware of to get good results without luminance. I have around 10 pixelmath expressions I vary from depending on the image (most picked up from other folks) and regardless of any pixelmath magic I still get better results with a lum channel added.

 

"Keep in mind, you need 4x as much data as you have, to get 2x the SNR... So if you have 11 hours now, you need around 44 hours to really get a reasonable improvement."

This one is interesting. Do you have data around this you can reference I'm curious why you'd need 4x data (not disagreeing with this, I have no clue).

 

EDIT: I just found this https://www.cloudyni...eturns-project/ which goes into this in more detail, super interesting. 

 

Thanks for the response. 


Edited by stobiewankenobi, 30 November 2020 - 09:54 PM.

  • mgstahl likes this

#4 imtl

imtl

    Vanguard

  • *****
  • Moderators
  • Posts: 2,140
  • Joined: 07 Jun 2016
  • Loc: Down in a hole

Posted 30 November 2020 - 10:15 PM

If I may, I think Jon might have read it incorrectly regarding the Lum. I read it like that at first but then realized that you might have meant something else.

 

Do you mean creating a luminance layer by integrating together Ha, OIII and SII (Or just Ha or whatever) or do you mean using L filter to add as luminance to combine as luminance channel?

 

If its the second, which I believe that was what Jon was referring to, then don't do that since it defies the whole purpose of the extremely high contrast of using NB filters for emission nebulae.

 

You create a NB image by combining the different NB channels in pixelmath or whatever you want to use and then start working on the blend. I separate the luminance NB layer and work on that and also create a tone map separately and then eventually combine both.

 

In case I was wrong with what Jon or you meant then I apologize in advance.


  • stobiewankenobi likes this

#5 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 10:19 PM

If I may, I think Jon might have read it incorrectly regarding the Lum. I read it like that at first but then realized that you might have meant something else.

Do you mean creating a luminance layer by integrating together Ha, OIII and SII (Or just Ha or whatever) or do you mean using L filter to add as luminance to combine as luminance channel?

If its the second, which I believe that was what Jon was referring to, then don't do that since it defines the whole purpose of the extremely high contrast of using NB filters for emission nebulae.

You create a NB image by combining the different NB channels in pixelmath or whatever you want to use and then start working on the blend. I separate the luminance NB layer and work on that and also create a tone map separately and then eventually combine both.

In case I was wrong with what Jon or you meant then I apologize in advance.


Hey there! Yes I mean creating a lum layer from Ha, or a combination of ha, oiii, and sii. I don't use lum or broadband filters at all when shooting nb for sure.

#6 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,718
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 30 November 2020 - 10:23 PM

Yeah I'm just not sure I agree. I have processed both with and without a luminance channel added in, and when adding a luminance channel I consistently get cleaner results. Additionally most of the bigger astro people I follow are doing this as well (like ak_astro just as an example who thoroughly documents their PI flows and get insane results) with narrowband imaging mono imaging. I get excellent contrast with an added luminance channel, the only real difference I see between my lum and non lum processing is more noise in the final image. 

 

All this being said it's totally possible there is some PixInsight magic I am not aware of to get good results without luminance. I have around 10 pixelmath expressions I vary from depending on the image (most picked up from other folks) and regardless of any pixelmath magic I still get better results with a lum channel added.

 

"Keep in mind, you need 4x as much data as you have, to get 2x the SNR... So if you have 11 hours now, you need around 44 hours to really get a reasonable improvement."

This one is interesting. Do you have data around this you can reference I'm curious why you'd need 4x data (not disagreeing with this, I have no clue).

 

EDIT: I just found this https://www.cloudyni...eturns-project/ which goes into this in more detail, super interesting. 

 

Thanks for the response. 

 

I know a lot of people combine lum with their NB. Either real lum, or synthetic lum. Just because a lot of people do something, doesn't mean its best. ;) "Cleaner results" is a highly subjective term. Contrast is a bit more specific. You will have higher contrast without blending a lum, than with. More specifically, you will have higher COLOR contrast...the colors of OIII will contrast better than the colors of Ha and SII. If you blend a lum, even one created by combining all the channels, you are now cross-contaminating the signals, and you LOSE that contrast. 

 

 

Regarding SNR. Signal grows linearly. Noise grows in quadrature. If you get 4x as much data. You have 4x as much signal, but 2x as much noise...so, 4/2 = 2...you get 2x the SNR. 

 

I always get good results for my NB images without blending any kind of luminance. That said, I also get 8-15 hours per channel, too...

 

VfCCOCg.jpg

 

igpvKdY.jpg


  • moonrider and mgstahl like this

#7 imtl

imtl

    Vanguard

  • *****
  • Moderators
  • Posts: 2,140
  • Joined: 07 Jun 2016
  • Loc: Down in a hole

Posted 30 November 2020 - 10:28 PM

Hey there! Yes I mean creating a lum layer from Ha, or a combination of ha, oiii, and sii. I don't use lum or broadband filters at all when shooting nb for sure.

Yes that is what I figured. But the way you wrote it was very LRGB jargon. The discussion with NB imaging usually is not involving the L to color ratio and all that. This really relates to broadband. The color in NB imaging comes from the NB. So you could definitely mix up the NB channels differently for the tone map (color) and another way for the "luminance channel". NB imaging is different to broadband and I think that was the confusing part in your original post.


  • Applal likes this

#8 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 24,718
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 30 November 2020 - 10:38 PM

I get excellent contrast with an added luminance channel, the only real difference I see between my lum and non lum processing is more noise in the final image. 

 

Do you actually have an example of this? I'd be curious to see... 

 

If you are creating a synthetic luminance, then you are in fact adding no additional signal. So if you are getting better results with a synthetic lum, it really would have nothing to do with the data, and more to do with what you are doing with the data. If you can process the individual channels properly, there is no reason you shouldn't be able to get the same quality, same apparent noise, etc. without spending any time on a synthetic luminance. 

 

A luminance channel, with LRGB, can be valuable because it is real, additional, distinct SIGNAL that truly adds to the total signal you have. RGB or NB, a synthetic lum doesn't actually add anything. So it all boils down to processing. In the end, if you get better results with a synthetic luminance, it has entirely to do with how you are processing it all...and nothing to do with the signals you have. That means the same quality could be achieved without the synthetic luminance...you may just have to approach how you handle the individual channels differently. 



#9 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 10:58 PM



Do you actually have an example of this? I'd be curious to see... 

 

If you are creating a synthetic luminance, then you are in fact adding no additional signal. So if you are getting better results with a synthetic lum, it really would have nothing to do with the data, and more to do with what you are doing with the data. If you can process the individual channels properly, there is no reason you shouldn't be able to get the same quality, same apparent noise, etc. without spending any time on a synthetic luminance. 

 

A luminance channel, with LRGB, can be valuable because it is real, additional, distinct SIGNAL that truly adds to the total signal you have. RGB or NB, a synthetic lum doesn't actually add anything. So it all boils down to processing. In the end, if you get better results with a synthetic luminance, it has entirely to do with how you are processing it all...and nothing to do with the signals you have. That means the same quality could be achieved without the synthetic luminance...you may just have to approach how you handle the individual channels differently. 

Yeah sure, I can actually provide an example of an image I finished recently and one that I can see you did as well on your Astrobin, and we both have similar total integration time. 

 

My image of Melotte 15

get.jpg?insecure

 

 

Your image of Melotte 15

get.jpg?insecure

 

I think I had around 17h of integration and you had 15h in yours. I think I had pretty good results in my final image. 

 

 

Another example with HOO with a Luminance blended OIII & Ha applied 

gallery_308856_12098_579638.png
 
 
 
 
 
A final note is that I am not super experienced at Astrophotography. So I am basing a lot of my discussion on my anecdotal experiences learning. I started this hobby about 1.5 years ago so I'm still learning quite a lot. 

Edited by stobiewankenobi, 30 November 2020 - 11:09 PM.

  • moonrider likes this

#10 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 11:12 PM

This has been an interesting convo so far, but I am still not sure if I have a good answer (whether lum is necessary or not). 

 

If you are shooting Narrowband, and you use Luminance in your processing (using Ha or comb narrowband NOT broadband or lum filter), once you have 4-5h of color data, is it better to push for 30+ hours of lum? Or do you need more color data as you scale your lum data? What should the ratios be for this type of processing?



#11 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 11:19 PM

If you are creating a synthetic luminance, then you are in fact adding no additional signal. So if you are getting better results with a synthetic lum, it really would have nothing to do with the data, and more to do with what you are doing with the data. If you can process the individual channels properly, there is no reason you shouldn't be able to get the same quality, same apparent noise, etc. without spending any time on a synthetic luminance. 

Regarding your statement here. Is this correct? I am not adding signal for sure, but with a luminance channel/mask added, I am masking the noise of the individual color channels with a cleaner luminance channel (especially if something like Ha, since Ha has nearly no noise in my imaging vs something like OIII)

 

When I am processing images, if I combine all narrowband data, and then split channels to RGB and re-combine with LRGB combine, I notice immediate noise reduction across the image, and no noticeable difference in contrast, and if there is a difference, it can easily be processed out, but I am left with a near noiseless image. 

 

Sara Wager has some good examples as well of this. 

 

Some other topics around this with Goofi chiming in on using Ha as Luminance here https://www.cloudyni...yer-narrowband/

 

Another great example here https://www.cloudyni...ing-melotte-15/


Edited by stobiewankenobi, 30 November 2020 - 11:23 PM.


#12 Huangdi

Huangdi

    Viking 1

  • *****
  • Posts: 963
  • Joined: 24 Jul 2019

Posted 30 November 2020 - 11:20 PM

This has been an interesting convo so far, but I am still not sure if I have a good answer (whether lum is necessary or not).

If you are shooting Narrowband, and you use Luminance in your processing (using Ha or comb narrowband NOT broadband or lum filter), once you have 4-5h of color data, is it better to push for 30+ hours of lum? Or do you need more color data as you scale your lum data? What should the ratios be for this type of processing?


I don't understand what you are saying. You say that you process your narrowband images LRGB. You do that by combining S, H and O into a master luminance and use the rgb blend from SHO as color data. You are not capturing luminance you're just making it from the sho data you capture so what's the question, really?

Unless you start shooting through a multi band filter with your mono cam (which would be true narrowband luminance), you can only increase the SNR of your pseudo Lum by increasing the total integration time.

That being said, I too like creating a Lum image from my 3 channels, but usually, to get clean signal, you should do 30 hours anyway. Narrowband signal can be really faint, especially in O3 and S2, just try to get as much as you can get.

#13 StephenW

StephenW

    Apollo

  • -----
  • Posts: 1,311
  • Joined: 04 Feb 2005
  • Loc: San Jose, CA

Posted 30 November 2020 - 11:23 PM

Adding back in Ha or some other channel as a "fake" Lum layer doesn't add any new signal, but it can greatly impact the presentation of the data in the final image

 

A good example is shown at the 29:10 mark in this tutorial:  https://www.youtube....h?v=j5zsZbHXrFE

 

Adding back in the Ha (fake Lum channel) doesn't add new signal but it does distribute the (strong) Ha signal across the RGB tone map in a manner that you don't get from just mapping SHO to RGB.  Personal preference will determine whether you like the result or not - personally I do :)

 

Could you achieve the same result by combining the Ha signal with S and O at an earlier stage in the processing pipeline?  Maybe, but not sure how you would do it though....



#14 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 11:31 PM

Adding back in Ha or some other channel as a "fake" Lum layer doesn't add any new signal, but it can greatly impact the presentation of the data in the final image

 

A good example is shown at the 29:10 mark in this tutorial:  https://www.youtube....h?v=j5zsZbHXrFE

 

Adding back in the Ha (fake Lum channel) doesn't add new signal but it does distribute the (strong) Ha signal across the RGB tone map in a manner that you don't get from just mapping SHO to RGB.  Personal preference will determine whether you like the result or not - personally I do smile.gif

 

Could you achieve the same result by combining the Ha signal with S and O at an earlier stage in the processing pipeline?  Maybe, but not sure how you would do it though....

Yeah this is exactly what I am doing and am referring to. 

 

I don't understand what you are saying. You say that you process your narrowband images LRGB. You do that by combining S, H and O into a master luminance and use the rgb blend from SHO as color data. You are not capturing luminance you're just making it from the sho data you capture so what's the question, really?

Unless you start shooting through a multi band filter with your mono cam (which would be true narrowband luminance), you can only increase the SNR of your pseudo Lum by increasing the total integration time.

That being said, I too like creating a Lum image from my 3 channels, but usually, to get clean signal, you should do 30 hours anyway. Narrowband signal can be really faint, especially in O3 and S2, just try to get as much as you can get.

See the last post here by StephenW. By adding a synthetic lum, or ha lum, to my SHO data, I get cleaner results.


Edited by stobiewankenobi, 30 November 2020 - 11:32 PM.


#15 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 11:37 PM

I don't understand what you are saying. You say that you process your narrowband images LRGB. You do that by combining S, H and O into a master luminance and use the rgb blend from SHO as color data. You are not capturing luminance you're just making it from the sho data you capture so what's the question, really?

Unless you start shooting through a multi band filter with your mono cam (which would be true narrowband luminance), you can only increase the SNR of your pseudo Lum by increasing the total integration time.

That being said, I too like creating a Lum image from my 3 channels, but usually, to get clean signal, you should do 30 hours anyway. Narrowband signal can be really faint, especially in O3 and S2, just try to get as much as you can get.

My question I suppose is: Should I increase my pseudo lum channel integration time? Or SII & OIII? Which is more effective at improving overall presentation of final image? I am mainly trying to understand if, once you have the detail from the color channels you want, should you solely focus on the "luminance" channel (in this case Ha)? Or is there a good ratio to keep in mind of luminance data acquisition vs color. In this case luminance == ha, color == ha, oiii, sii.


Edited by stobiewankenobi, 30 November 2020 - 11:38 PM.


#16 StephenW

StephenW

    Apollo

  • -----
  • Posts: 1,311
  • Joined: 04 Feb 2005
  • Loc: San Jose, CA

Posted 30 November 2020 - 11:43 PM

>once you have the detail from the color channels you want, should you solely focus on the "luminance" channel

 

In general, I would say "yes". 

 

If you are going to use a fake lum channel, then once you have sufficient color channels, getting more signal in the lum channel to improve structure/detail seems like the right investment.



#17 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 11:48 PM

>once you have the detail from the color channels you want, should you solely focus on the "luminance" channel

 

In general, I would say "yes". 

 

If you are going to use a fake lum channel, then once you have sufficient color channels, getting more signal in the lum channel to improve structure/detail seems like the right investment.

This makes sense to me as well, and it was my first intuition. I'm currently adding another 8h in Ha tonight to my Pacman under to this assumption so I'm glad another thinks it's the "right" thing lol. 

 

Appreciate it!


Edited by stobiewankenobi, 30 November 2020 - 11:48 PM.


#18 Huangdi

Huangdi

    Viking 1

  • *****
  • Posts: 963
  • Joined: 24 Jul 2019

Posted 30 November 2020 - 11:49 PM

My question I suppose is: Should I increase my pseudo lum channel integration time? Or SII & OIII? Which is more effective at improving overall presentation of final image? I am mainly trying to understand if, once you have the detail from the color channels you want, should you solely focus on the "luminance" channel (in this case Ha)? Or is there a good ratio to keep in mind of luminance data acquisition vs color. In this case luminance == ha, color == ha, oiii, sii.


This probably comes down to personal preference. Narrowband often is different to rgb imaging. There are many instances where you have very weak signals that require massive exposure times to even reveal them... So I'd say it depends on the target. I don't think that there is a clear answer to this.

#19 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 30 November 2020 - 11:53 PM

This probably comes down to personal preference. Narrowband often is different to rgb imaging. There are many instances where you have very weak signals that require massive exposure times to even reveal them... So I'd say it depends on the target. I don't think that there is a clear answer to this.

Yeah this makes sense for sure. That is why I specified that it'd be only once you have the detail from the color you want. For example there is a return at which there is no more OIII data etc... so pushing into Ha longer. 

 

This all makes sense to me.



#20 Peregrinatum

Peregrinatum

    Surveyor 1

  • *****
  • Posts: 1,560
  • Joined: 27 Dec 2018
  • Loc: South Central Valley, Ca

Posted 01 December 2020 - 12:10 AM

forget the time to capture Lum, just process an extracted lum and blend it after stretchting



#21 imtl

imtl

    Vanguard

  • *****
  • Moderators
  • Posts: 2,140
  • Joined: 07 Jun 2016
  • Loc: Down in a hole

Posted 01 December 2020 - 12:36 AM

I think you should stop calling it lum because people still think you are actually capturing lum.

If you just add and blend Ha back as luminence channel you are not gaining any signal, you are just changing the weights in the SHO blend. Since Ha is the strongest signal then no wonder you are getting ''cleaner'' results. But that you will get if you just up the weight of your Ha in your pixel math blend. Has nothing to do with luminance channel. Which is what others are trying to tell you as well I believe.
  • gatsbyiv likes this

#22 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 01 December 2020 - 12:46 AM

I think you should stop calling it lum because people still think you are actually capturing lum.

If you just add and blend Ha back as luminence channel you are not gaining any signal, you are just changing the weights in the SHO blend. Since Ha is the strongest signal then no wonder you are getting ''cleaner'' results. But that you will get if you just up the weight of your Ha in your pixel math blend. Has nothing to do with luminance channel. Which is what others are trying to tell you as well I believe.

Yeah that is possible. Every single reference I have seen for this calls it "lum/luminance"... As far as I'm aware Luminance is not something that is specific to a "filter" or similar like discussed. Luminance is just a channel or a layer that can be added to an image. Specifically Luminance is "The Y is known as the Luminance channel and it represents brightness within the image."

 

Example of folks using the term "luminance" when discussing specifically narrowband: https://www.flickr.c...ro/50276326027/

 

Along with most youtube video's about this when discussing narrowband processing, or pixinsight flows, call it lum/luminance.

 

But it's a fair assessment that it might be confusing people, but it is fairly common when discussing PixInsight processing workflows (like tonemapping) from what I have seen. Every guide/forum/discussion around this calls it luminance.

 

 

 

In any case, I am curious if you have an example pixelmath expression that would sufficiently do what I am doing here, without blowing out the red channel. Adding Ha as Lum with something like LRGB combine adds it as..luminance, not as a blended color in Pixelmath (albeit I'm no pixelmath pro). 

 

Would be curious if you have example expressions that do this. I have tried a good few (like these https://thecoldestni...th-pixelmath/) 

 

And I definitely don't get similar results as when I blend with luminance. 

 

Appreciate all the dialogue here. 


Edited by stobiewankenobi, 01 December 2020 - 12:50 AM.


#23 imtl

imtl

    Vanguard

  • *****
  • Moderators
  • Posts: 2,140
  • Joined: 07 Jun 2016
  • Loc: Down in a hole

Posted 01 December 2020 - 01:07 AM

So just looking at the flickr example you gave it to show exactly what we all wrote about luminance in narrow band. The person just took Ha as his/hers luminance. So basically only the tone map is SHO. The details come from Ha. That is NOT a Hubble palette in the sense of what Jon and others are saying. You only have some info from OIII and SII carved in your color and not in the details. Which is what's important.

 

By doing this, you do not show the SHO structure, you're just showing Ha structure colored with SHO tone map.

The fact that it comes out cleaner is because you're using your strongest signal for structure and details, which is Ha. When you do a full SHO blend then you mix the Ha signal with weaker (noisier most of the time) OIII and even weaker SII. So you will get noisier image. But you would also get much richer structural details because now you have all three Ha, OIII and SII in the mix. And not just in the tone map. Which is what you eventually want in SHO image. 

 

I hope I am making myself clear :O

 

What you are doing is:

 

Luminance: Ha

Tone map: R= S; G=Ha; B=OIII.

And then LRGBcombine.

 

What Jon and others are talking about is to use pixel math for example and do

 

R: SII*0.6+Ha*0.4

G: OIII*0.4+Ha*0.3+SII*0.3

B: OIII

 

That is one blend I use. Or just simple non-modified SHO:

 

R=SII

G=Ha

B=OIII

 

This gives you:

Luminance AND tone map to be of the same blend. What's important here is not so much the tone map but more the "luminance" because now you actually have details from ALL three NB wavelengths.



#24 stobiewankenobi

stobiewankenobi

    Mariner 2

  • -----
  • topic starter
  • Posts: 221
  • Joined: 25 Apr 2019
  • Loc: Laguna Beach, CA

Posted 01 December 2020 - 01:13 AM

So just looking at the flickr example you gave it to show exactly what we all wrote about luminance in narrow band. The person just took Ha as his/hers luminance. So basically only the tone map is SHO. The details come from Ha. That is NOT a Hubble palette in the sense of what Jon and others are saying. You only have some info from OIII and SII carved in your color and not in the details. Which is what's important.

By doing this, you do not show the SHO structure, you're just showing Ha structure colored with SHO tone map.
The fact that it comes out cleaner is because you're using your strongest signal for structure and details, which is Ha. When you do a full SHO blend then you mix the Ha signal with weaker (noisier most of the time) OIII and even weaker SII. So you will get noisier image. But you would also get much richer structural details because now you have all three Ha, OIII and SII in the mix. And not just in the tone map. Which is what you eventually want in SHO image.

I hope I am making myself clear :O

What you are doing is:

Luminance: Ha
Tone map: R= S; G=Ha; B=OIII.
And then LRGBcombine.

What Jon and others are talking about is to use pixel math for example and do

R: SII*0.6+Ha*0.4
G: OIII*0.4+Ha*0.3+SII*0.3
B: OIII

That is one blend I use. Or just simple non-modified SHO:

R=SII
G=Ha
B=OIII

This gives you:
Luminance AND tone map to be of the same blend. What's important here is not so much the tone map but more the "luminance" because now you actually have details from ALL three NB wavelengths.

I think I understand yeah this was something I was not aware of. So doing ha lum with oiii and sii is basically just color coding ha data. I Definitely didn't know that.

Interesting. I hadn't noticed a loss of detail or structure data with any of my Narrowband images as of yet.

Do you have an example of a DSO with lots of structure data from sii of oiii? From the few dozen I've imaged so far in my noob journey I notice most of it comes from ha and mostly blob data without many hard lines comes from oiii. Sii I have noticed more hard lines though.

I'll have to pay more attention to this this is super helpful. Thank you for explaining!

Edited by stobiewankenobi, 01 December 2020 - 01:14 AM.


#25 imtl

imtl

    Vanguard

  • *****
  • Moderators
  • Posts: 2,140
  • Joined: 07 Jun 2016
  • Loc: Down in a hole

Posted 01 December 2020 - 01:17 AM

I think I understand yeah this was something I was not aware of. So doing ha lum with oiii and sii is basically just color coding ha data. I Definitely didn't know that.

Interesting. I hadn't noticed a loss of detail or structure data with any of my Narrowband images as of yet.

Do you have an example of a DSO with lots of structure data from sii of oiii? From the few dozen I've imaged so far in my noob journey I notice most of it comes from ha and mostly blob data without many hard lines comes from oiii. Sii I have noticed more hard lines though.

I'll have to pay more attention to this this is super helpful. Thank you for explaining!

Its because usually Ha is very dominant. And getting OIII and especially SII to pop out requires a lot of integration time, or noisy results which people hide using....Ha as luminance :)

 

This thread, which I really liked but for some reason did not pick up as I thought it would, gives you some insight on the relative strength between SHO channels. I think if people keep adding their own comparisons for different NB objects then we could build a really nice and informative data base for others to use.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics