Share:
Notifications
Clear all

2022-05-29: APP 2.0.0-beta2 has been released !

Release notes

Download links per platform:

windows 2.0.0-beta2

macOS x86_64 2.0.0-beta2

macOS arm64 M1 2.0.0-beta2

Linux DEB 2.0.0-beta2

Linux RPM 2.0.0-beta2

[Sticky] Ha + OIII Dual Narrowband Workflow

Page 1 / 2

(@shedastro)
White Dwarf Customer
Joined: 2 months ago
Posts: 16
Topic starter  

Hello

would it be possible to have an unlinked stretch option for OSC data, so that when shooting with L-eXtreme filters we can get the narrowband colours right away rather than the long winded way we have to do it now…the option is in PI, which I used to use years ago, and saves such a lot of time when using these filters….

or is the option there now, and I am missing it….

cheers 😀


ReplyQuote
(@shedastro)
White Dwarf Customer
Joined: 2 months ago
Posts: 16
Topic starter  

I guess that’s a no then….☹️


ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4856
 

We're not always able to reply within a day, sorry about that.

Not sure what that process exactly is in PI, but it sounds a lot like extracting the Ha and OIII directly? This can be choosen in tab 0, the algorithm option.


ReplyQuote
(@shedastro)
White Dwarf Customer
Joined: 2 months ago
Posts: 16
Topic starter  

@vincent-mod 

no I know how to extract them, but is an unnecessary process if you had an unlinked stretch option….

I mean when stretching in APP the stretch applied is linked and is auto, and all the colours are stretched to a specific range preset in the software, with green dominant, but what about an unlinked stretch where it just stretches the RGB all the same amount which with the mentioned filters produces an HOO image right away, no separate extraction needed….it’s much simpler….


ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4856
 

I asked Mabula and he says it's better to perform a linear stretch via the RGBCombine tool, which would work better than an unlinked stretch. I tend to use it like that, combined with extracting the narrowband data first as that produces a cleaner result (usually less noise). We don't have an unlinked stretch indeed.


ReplyQuote
(@shedastro)
White Dwarf Customer
Joined: 2 months ago
Posts: 16
Topic starter  

@vincent-mod 

well I am sorry but it does not work, it just messes up my image, the unlinked stretch in PI works wonders with these duo band filters, APP does not work well with them at all, it’s a shame, as I like APP, but it just does not produce the correct colours at all in the final image….☹️


ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4856
 

That's a pitty, would you mind sharing your dataset? I can try to see if I can get better colors and then share my workflow.


ReplyQuote
(@shedastro)
White Dwarf Customer
Joined: 2 months ago
Posts: 16
Topic starter  
Posted by: @vincent-mod

That's a pitty, would you mind sharing your dataset? I can try to see if I can get better colors and then share my workflow.

Here is a link to the stacked fits file, I then put this in PI and just hit unlinked stretch and get the image shown here….no other processing needed….not sure how I could get that with APP…

https://www.dropbox.com/s/b66txsef1m0pwpv/Rosette-RGB-session_1-lpc-cbg.fits?dl=0

CD238AAD C007 4095 B86E F6E262D554CB

ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4856
 

That's totally possible, but I would need some data to be able to show you how to do that nicely. If you could upload 5-10 subs of your data with your master calibration files, that should be enough; I'll also check your final fits to see what APP shows.


BillyBoy liked
ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4856
 

Ok, checked out your integration. It seems this is a RGB integrated result, so a stretch of that results in a red nebula. Also in PI when I do this unlinked:

image

So I guess you did some other steps first in either APP or PI? Could you give those?

I do understand now what you're after, in the sense that unstretched would give you an "optimal" stretch per channel, which changes the color (thanks to Wouter for filling me in there 🙂 ). That will however introduce more noise from the green channel for instance. I think this is why Mabula hasn't implemented this kind of stretch standard. I would start with a Ha and OIII integration from the start in APP, which is also almost automatically, as that will extract the Ha very nicely with minimal noise added to it. You then get both mono integrations and can very nicely combine those. It's maybe not a 1-click step, but does result in a nice overall result.


BillyBoy liked
ReplyQuote
(@shedastro)
White Dwarf Customer
Joined: 2 months ago
Posts: 16
Topic starter  

@vincent-mod 

hmmm, that is odd, as it was the same fits file put into PI with an unlinked stretch that gave the image I posted above….🤔

so it would be easier for me if you could just give me the workflow in APP and I will try with my data, that would be easier than uploading it, as I don’t have the master calibration files, only the RAW files, not sure why but can’t find them…

cheers


ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 5 years ago
Posts: 3159
 

Hi BillyBoy @shedastro,

The best and most optimal workflow can only be accomplished on dual narrowband data if you provide the Ha and O3 stacks as created with the Ha-O3 extract Ha/O3 demosaic algorithms.

Then you only need to create a bi-color with the RGB Combine Tool, select the HOO1 or HOO2 formula and start tweaking the multipliers a bit, you will soon have a beautiful bi-color composite.

Since you have only provided a RGB stack, I will try to process this now and will report back later... did you create the RGB stack with Ha-O3 color ? or the default in 0) RAW/FITS? If you used the default, the data is already less optimal than could be, so therefor I ask 😉

Mabula


ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 5 years ago
Posts: 3159
 
Posted by: @shedastro
Posted by: @vincent-mod

That's a pitty, would you mind sharing your dataset? I can try to see if I can get better colors and then share my workflow.

Here is a link to the stacked fits file, I then put this in PI and just hit unlinked stretch and get the image shown here….no other processing needed….not sure how I could get that with APP…

https://www.dropbox.com/s/b66txsef1m0pwpv/Rosette-RGB-session_1-lpc-cbg.fits?dl=0

CD238AAD C007 4095 B86E F6E262D554CB

BillyBoy @shedastro,

hmmm, that is odd, as it was the same fits file put into PI with an unlinked stretch that gave the image I posted above….🤔

If I do an unlinked stretch on the provided data in Pixinsight I get the same result that Vincent showed, so not that odd I think, that image of yours has had much more work done than only an unlinked stretch. For starters, it has a clear and aggressive contrast boost, looking at the details in the nebula and the colors...

Maybe you can show us a screenshot of how you would get such a result with an unlinked stretch in Pixinsight actually showing it is just 1 action on the exact same data? Your current screenshot leaves no indication at all that it is done like that.

Furthermore, you argue that an unlinked stretch would be the solution to always get good color on such narrowband objects? That will simply not be, it really depends on the object characteristics if such an unlinked stretch would look nice or not, at least in my own experience.

There is no set formula to always get a good result from a narrowband composite because it will always depend on:

  • the quality of both Ha and O3 data in this case
  • the object characteristics meaning how the nebula composition really is per object in which part of the nebula and what emmissions actually are there.

So a change in this between datasets will always need different formula to get the most pleasing result or even to get similar results, different formula will be needed.

Anyhow, I am done with processing your file and will show in a next psot how I would process that file.

Mabula

 

 

 


BillyBoy liked
ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 5 years ago
Posts: 3159
 

Processing by Mabula on the data provided in BillyBoy's dropbox link

Data provided:

DataProvided

 

First step, I load your RGB stack into the RGB Combine Tool. I am assuming that it is in fact a Ha-O3 color integration.

So in the RGB Combine Tool, the RGB stack is automatically split into R,G,B channels and you can indicate the formula and which filter was actually used.

So I select the HOO1 formula, and the first channel is assigned to H-alpha and the 2nd and 3rd both to Oxygen III.

Calculate with HOO1 formula gives this, this is almost identical to the unlinked stretch in Pixinsight that both Vincent and me get but now using a linear color ratio change since RGB COmbine is doing linear operations on the linear original data (which keeps better control !) and then a preview stretch (which is identical for all channels) with APP's previewfilter:

HOO1 formula
HOO1

Pixinsight unlinked stretch is almost identical, hard to see how it could ever result in the result that you posted BillyBoy @shedastro. Maybe you need to check what actually happened to get to your result and show us stept by step?:

unlinked stretch pixinsight

Then next step, let's check HOO2 formula in APP, set the formula to HOO2 and click new formula to apply that formula to your composite and then again click (re-)calculate:

HOO2 formula
HOO2

The HOO2 result is a bit bleak, on most datasets it will actually look better than HOO1. Maybe because we are not working on the Ha and O3 stacks, but on a RGB stack as provided, that is hard to tell at this point.

Anyway, with a bit of experience you will use the HOO1 and HOO2 formula as starting points. You need to tweak the multipliers and color ratios in such a way that you get the colors that you prefer in such a false color narrowband image. In this case, I am simply trying to produce the result as shown by BillyBoy with a clear blue interior of the Rosette and a yellowish exterior of the nebula.

Notice that i use the exact same slider values for the Oxygen III parts of the original RGB compositie, which are

lpc-cbg.fits-2-Oxygen III  <- -> 2 denotes it was the GREEN channel in the original RGB

lpc-cbg.fits-3-Oxygen III <- -> 3 denotes it was Blue

Essential with Bi-Colors like these is that you need to realize that the Oxygen III (or O3) signal is much weaker than the H-alpha signal. So we boost O3 with setting mulitplier to 3.25 which leads to similar colors as what we are aiming for.

I made small adjustments in the HOO2 formula and I used the H-alpha layer to dominate luminance for the composite since that signal has much higher quality.

In the preview fitler, I enabled color saturation and I see this:

Custom formula
Custom

Further processing is in the next post...


ReplyQuote
(@shedastro)
White Dwarf Customer
Joined: 2 months ago
Posts: 16
Topic starter  

@mabula-admin 

thanks very much for this, I am starting to understand this a bit more, it’s knowing how to use the sliders, and what and where to move them, to get the correct colour, is this just trial and error, or are there set values for set colour combinations of dual narrowband data sets….?


ReplyQuote
(@shedastro)
White Dwarf Customer
Joined: 2 months ago
Posts: 16
Topic starter  

So from the original L-eXtrme data, how would I get the Ha and OIII individual stacks, when I first stack the data what do I choose, and I am talking way before I use the combine tool, do I stick with a certain formula, instead of just assigning the stack to RGB…??


ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 5 years ago
Posts: 3159
 

I saved the RGB Combine Result and first tweaked a bit with the preview filter for the B(lack) point, the ST(retch), CON(trast) and color saturation :

Custom previewfilter adjustments saturation contrast

Clearly a big contrast boost is needed to make it look stand out more and get more nebula details showing nicely

Final adjustment then can be nicely done with the HSL Selective Color tool to get more blues and have yellow stand out more, and kill color noise in the background:

HSL selective color

1) boost yellow, select yellow, background - 50% of data range (check with show color selection which pixels are selected), set R(ed)<-->CY(an) slider to -0,75 to add RED to yellow pixels and set B(lue)<-->YE(llow) to 0,75 to add more yellow to the selected yellow pixels. Set saturation to 050 to boost saturation of those pixels. Notice that I disabled the auto DDP stretch to keep my preview filter tweaks intact going forward with processing. Check the adjustment with calculate current adjustments, keep the adjustment with keep current adjustments  to start work on other pixels:

Custom SelectiveColor 1

2) boost blue, select blue, background -25% data range, set B -YE to -0,8 to add Blue and set G(reen)-MA(genta) to -0,4. This makes the blue pixels more blue and prevents a magenta cast since those pixels also still have weak red parts as well. Also more saturation 050. To protect the bright blue stars, you can lower the data range of the selected pixels by lowering the HIGH slider. I lowered it to only 0,03

Step 1 and 2 could be repeated again and again and adjusted until it pleases you. This is very subjective with narrowband, there is no good or bad here, good is what you like, because the nebula in broadband RGB is simply mostly RED so this is not how this nebula looks if you would fly in space and see it from your rocket window 🙂

3) Kill background color noise, select ALL pixels from 0-background + 1*noise and check the selected pixels with the show color selection button :

Background noise Pixels

The weak H-alpha emmisions will not be harmed by this selection as can be seen in the mask :-). Those faint signals are visually above the noise level and that is represented by the background + 1*noise threshold. Simply set saturation to -100 which kills the color noise on the selected pixels.

Custom SelectiveColor 3

 

4) Now I  decided to manipulate the red pixels again by aggressively adding red and yellow strongly, R-CY -0,90 and B-YE +0,70 and SAT 050:

Custom SelectiveColor 4

Think this more or less like the result of BillyBoy which i tried to re-create in APP. (But since it is NarrowBand, I could have gone for another color composition of course.)

This post was modified 2 months ago by Mabula-Admin

ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 5 years ago
Posts: 3159
 

To finish, I used the Star Reducer in APP with

correction radius 1.2

correct star halo 1.0

star size 65%

peak intensity 65%

noise threshold K=1 (means all stars above the noise floor are reduced)

Final Star Reducer

ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 5 years ago
Posts: 3159
 
Posted by: @shedastro

@mabula-admin 

thanks very much for this, I am starting to understand this a bit more, it’s knowing how to use the sliders, and what and where to move them, to get the correct colour, is this just trial and error, or are there set values for set colour combinations of dual narrowband data sets….?

Excellent, that is very nice to hear 🙂 For these RGB Combine formula, it does take a bit of experience I would say. The formula presets are there to give you a starting point to make your own custom formula. But same principle apply, boost O3 versus Ha when Ha is much better quality or clearer signal for instance.

Like i indicated before, the formula that works best for you will always depend on the quality of the data in general and the object that you targeted. So there is no magic formula that will always work on each dataset on each object.

 

 


BillyBoy liked
ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 5 years ago
Posts: 3159
 
Posted by: @shedastro

So from the original L-eXtrme data, how would I get the Ha and OIII individual stacks, when I first stack the data what do I choose, and I am talking way before I use the combine tool, do I stick with a certain formula, instead of just assigning the stack to RGB…??

In 0) Raw/FITS, you would get the H-alpha stack by setting the demosaic algorithm to Ha-O3 extract Ha 😉 . For O3, you use Ha-)3 extract O3.

Then those 2 stacks are the most optimal to use for Ha-O3 bicolor composite. You would load the Ha and O3 stacks into RGB Combine and again choose the HOO1 or HOO2 formula as a starting point 😉

 


BillyBoy liked
ReplyQuote
Page 1 / 2
Share: