[Sticky] Combining R, G, B with Ha & OIII data using an Optolong L-eNhance filter

Hi Dale @dalemschultz,

Do you still need an aswer to your question? Please accept my apologies for the late answer to your question.

I'm a little late to this party, but I too want to understand better how to implement NB filters in post processing. Currently, I am getting out of memory errors. My system only has 16GB of RAM and that's not enough to even combine any two of the integrated files, let alone 6 if I want to combine R,G,B,Ha & OIII twice. Something I'm missing, besides more RAM?

More than 16GB of RAM would be needed if your images that you want to combine in the RGB Combine Tool are rather large in pixel dimensions to start with. Was this the case?

Perhaps you can try with APP 2.0.0-beta32 and let me know if all is okay?

Mabula

Hi Josh @ibid,

That separation between Ha and O3 will work for all the dual-band filters that would record H-alpha and O-3 signals. So yes, it should definitely work fine on the Vaonis filter.

Please let me know if all is clear, thanks!

Mabula

@mabula-admin I’ll bet this is the first you will have heard this excuse: I had a heart transplant last night and therefore will not be able to report back anytime soon. However, I’m thrilled to hear that the functionality exists and I look forward to using it. Thank you!

Hi Josh @ibid,

I wish you the very best of health and strength in your recovery ! You are most welcome.

Hi again. I promised a long time ago that I would give an update after trying various processing methods with my L-Pro filter. As suspected by Mabula, I found it best to simply treat it as a broadband filter and integrate shots taken with it the same as with e.g. a UV/IR filter. Techniques such as described in this post for L-Enhance and similar filters seem to just complicate the workflow with no better result when using the L-Pro. 

However, I have now also bought an L-Para filter for my ASI2600 Duo OSC camera. First target with it was NGC4244- Silver Needle galaxy. But when I processed the exposures, I was unable to Register the H-Alpha exposures following the guidelines given. The OIII part worked fine, but the H-alpha seemed to have too few stars to be Registered together with the OIII and RGB exposures even though the H-Alpha and OIII were the same exposures. I tried various settings in APP but gave up as none seemed to help. That said, I struggle to understand some of the help descriptions given in APP! Is there something I have likely missed in the process, or is it simply a fact of life that a low H-alpha shot simply won't be registrable? NB, they were 300sec exposures at f6.3 with a 9.25 SCT in a Bortle 3 zone and no moon, so plenty of stars were visible in the FIT files prior to processing...

Thanks!

Hi Mark @mfc,

I promised a long time ago that I would give an update after trying various processing methods with my L-Pro filter. As suspected by Mabula, I found it best to simply treat it as a broadband filter and integrate shots taken with it the same as with e.g. a UV/IR filter. Techniques such as described in this post for L-Enhance and similar filters seem to just complicate the workflow with no better result when using the L-Pro. 

Great, it should be like that actually. The L-Pro is more a broadband than a narrowband filter since it only suppresses light waves related to light-pollution, right?

However, I have now also bought an L-Para filter for my ASI2600 Duo OSC camera. First target with it was NGC4244- Silver Needle galaxy. But when I processed the exposures, I was unable to Register the H-Alpha exposures following the guidelines given. The OIII part worked fine, but the H-alpha seemed to have too few stars to be Registered together with the OIII and RGB exposures even though the H-Alpha and OIII were the same exposures. I tried various settings in APP but gave up as none seemed to help. That said, I struggle to understand some of the help descriptions given in APP! Is there something I have likely missed in the process, or is it simply a fact of life that a low H-alpha shot simply won't be registrable? NB, they were 300sec exposures at f6.3 with a 9.25 SCT in a Bortle 3 zone and no moon, so plenty of stars were visible in the FIT files prior to processing...

That sounds like a problem with your H-alpha files then, normally it will be very straight forward to analyze and register the H-alpha data. Problems are much more likely to occur with SII or O3 data because that signal is usually much weaker than H-alpha. With 300 seconds exposure at F6.3 I would expect no problems as well with H-alpha data.

Did you double check your calibration files for H-alpha? What happens in analyse stars without calibration data? If the H-alpha data calibrated still has clear noise/hot pixels, maybe APP is locking onto hot pixels for stars? Did you try enabling the cosmic ray/noise reducer? If not, try that and set the lower limit area size to 15 and try again.

Please let me know if this helps 😉

Mabula

Hi, and thanks Mabula - using Cosmic Radiation/NR = 15 worked! The number of stars dropped from 500+ to around 200, so I selected Registration using triangles and Normal / Projective mode. The end result is much clearer with quite visible stars. Once that was proven, I tried again using my calibration frames but retaining CR/NR=15 and the result was better again. So I assume it was background noise, and not the calibration frames which were the issue. I also re-ran the same change on the OIII files and got a far cleaner result. I am sure there won't be an issue to combine the various channels now.

My first question is why? I was surprised how noisy the L-Para shots were - could that possibly be caused by Aurora activity (I am 61 deg Nth and I am pretty sure the Aurora was out then)? Shots on SH2-129 with the L-Para taken the night prior were also very noisy- even though they were at f2.2 using a hyperstar. Do you have other suggestions, or is this "normal"? Or maybe orientation of the L-Para filter is important (despite online posts saying it 'shouldn't' be. In my books, "shouldn't" isn't the same as "doesn't"). I may reverse it next trip to my observatory, to see if it affects the results...

I will now continue the process described in this chain, but have some other qns before I do: 

• Why is it recommended to enable Dynamic Distortion correction and disable Same camera and Optics, if all the shots are taken with the same configuration, albeit different filters?
• As all my shots are with OSC, using either UV/IR or L-Para filters, what would you recommend as the best Formula- RGBHOO, or something else?
• In the description, it recommends adding the OIII as both Blue and Green. Why? Could I instead use the sliders to push both B & G to 100%?
• What do the "bg 1.000" & "L" sliders do? I played with them but couldn't work out what was happening, except that the result quickly became terrible!

Thanks again, Mark

Hi Mark @mfc,

using Cosmic Radiation/NR = 15 worked! The number of stars dropped from 500+ to around 200, so I selected Registration using triangles and Normal / Projective mode. The end result is much clearer with quite visible stars. Once that was proven, I tried again using my calibration frames but retaining CR/NR=15 and the result was better again. So I assume it was background noise, and not the calibration frames which were the issue. I also re-ran the same change on the OIII files and got a far cleaner result. I am sure there won't be an issue to combine the various channels now.

Excellent !

My first question is why? I was surprised how noisy the L-Para shots were - could that possibly be caused by Aurora activity (I am 61 deg Nth and I am pretty sure the Aurora was out then)? Shots on SH2-129 with the L-Para taken the night prior were also very noisy- even though they were at f2.2 using a hyperstar. Do you have other suggestions, or is this "normal"? Or maybe orientation of the L-Para filter is important (despite online posts saying it 'shouldn't' be. In my books, "shouldn't" isn't the same as "doesn't"). I may reverse it next trip to my observatory, to see if it affects the results...

I have no experience myself yet with the L-Para filter, maybe the gain is set too high or perhaps it is the sky conditions somehow.

• Why is it recommended to enable Dynamic Distortion correction and disable Same camera and Optics, if all the shots are taken with the same configuration, albeit different filters?

This is not recommended for regular integrations, only for registration of mosaic panels 😉 for regular integrations, by all means, keep same camera and optics enabled.

• As all my shots are with OSC, using either UV/IR or L-Para filters, what would you recommend as the best Formula- RGBHOO, or something else?

There is no such thing as a best fornula, it all depends on many factors actually. It depends on the object, it depends on the quality of the data, it depends on what you like. Ideally, you use one of the preset formula in the RGB Combine Tool as a starting point to make your own customized formula which will be most to your liking.

• In the description, it recommends adding the OIII as both Blue and Green. Why? Could I instead use the sliders to push both B & G to 100%?

The OIII signal has a wavelength just between green and blue. That is the natural color of OIII. But you can simply change the fornula yes, to make it look completely different. Alter the factors in the tool and push calculate to see how that affects things.

• What do the "bg 1.000" & "L" sliders do? I played with them but couldn't work out what was happening, except that the result quickly became terrible!

Bg will afffect the background level of that channel, normally it does not need tweaking since the tools normalization will make sure that the backgrounds are normalized properly. The L slider is for Luminance. So it enables you to use 1 or more of your channels for the composite to provide the Luminance layer of the result. So normally, you will either have a luminance layer shot with a luminance filter, or you use the channel that has the highest quality, then the composite will have less noise. If you only use 1 layer for L, be careful with setting it to 100% for L. Usually, it is better to keep it around 60-80% 😉

Let me know if this helps.

Mabula

@mabula-admin thanks again, I will continue to play with the settings as you suggested and get back to you and the others on this post with the results. 
M.

@mabula-admin Thanks Mabula, that worked. I still flounder around trying to work out what mix I need in order to get a good end result, but I have found some combinations that seem to be OK. I am sure someone with more experience and skill would do much better. I haven't though, worked out how to use Luminance. I tried re-adding the H-alpha channel as an L-layer, but even at 10% it simply whited out the result, so I gave up trying. One interesting thing I noticed is that if I didn't split the RGB stack initially and just tried to add it as a single file, APP automatically loaded it for me as separate R, G & B! That is nice and saved quite a few clicks in the processing...

Hi Mark, that sounds great !

Regarding the H-alpha layer as L, sometimes you need to tweak the normalization settings in the RGB Combine tool to get better results initially so please try that if you see such a white out result 😉

Indeed, RGB stacks/lights are directly splitted in the tool!

Mabula

Hi all ...
... and sorry for this beginner question.

I managed to process OSC data from my brand new ASI2600MC. Of course, still in beginner mode, but I am quite pleased and my M81/M82 look good, I even have "the cigar smoking" 😉

I now took 3 hours of duoband (6nm Ha OIII filter). I read in the Cloudy Nights forum that M82 Cigar Galaxy has some Ha, which will help even more to get the "smoking" part. Processed as OSC, it looks quite promising, even without further processing in Photoshop:

But I would like to do HaRGB processing and started to read some of the tutorials here.
My questions, before I start to process anything:

• When should I do the steps lpc/cbg (light pollution correction and background calibration)?
  Before or after extracting the R B G channels?
  Before or after Combine RGB
• Same question for csc (correct star colours).

Or in other words, in the basic work flow in post #1 ... where to flick in lpc and csc? And HSL colour - at the very end, or also in an earlier stage?

Thank you and best regards.

Michael 

What I would do, is the following:

Split the RGB Image into the separate channels

Split the NB image into RGB Channels, Discard the G and B Channels, just keep the R channel and rename it to Ha

Re-Construct the image with the R (Ha), R, G, B using the RGB tool using the HaRGB profile

Simon

Hi Simon.

Thank you - with this I made it.
I did the following:

1. RGB is my FITS from 3 session, drizzled, lpc/cbg, csc and a bit SC
   --> load as Light, in 2)Calibrate I tick "split channel" and SAVE Light Frames
   --> Result are 3 separate R, B, G FITS
2. RGB development of HaO3 duoband Lights, with Master-Flat,Dark,Bias and BPM, same drizzle, lpc/cbg but no csc, no SC
   --> As you wrote: Split into R,G,B and keep only R, I rename it to Ha
3. Load the separate R, G, B and Ha FITS, registered and "save registered" frames
4. Load Ha-reg --> sr, removes stars. I to have the star colours from my RGB integration.

This just to explain, what I did additionally:
- used a fully developed RGB
- did lpc on the HaO3 integration
- needed to register the 4 channels (because from various sessions, so different #pixels)
- and finally removed stars from the Ha channel (= red channel of RGB integration of HaO3 lights)

Question 1: Are all these steps (lpc, csc, SC, sr and registration) in the correct sequence?

-----

Then your last step ...

Posted by: @stastro

↑

Re-Construct the image with the R (Ha), R, G, B using the RGB tool using the HaRGB profile

... was (my) breakthrough. I think, my mistake was to not assign the Ha channel to Red.
The result is a bit too red, but this can be corrected in Photoshop. Main goal is reached: more "cigar glow" by adding the Ha.

Thank you 🙂

Hi Mabula @mabula-admin

What I do not understand: How can I use the Ha channel, created by "Ha-OIII extract Ha".

I always get weird colours, e.g.

Here it's HaRGB 1, Ha channel assigned to Red. ("Red" as filter selected)
I tried to assign it to Hydrogen Alpha (selected from the filters) ... comparable strange colours.

I tried some more, I played with the sliders ... and obviously have no clue. It's like the "ape clicking a keyboard".

==> Is there any tutorial, how to use Combine RGB and which slider to use to achieve what?

Thank you and best regards.

Michael 

Posted by: @michaelacg

↑

Question 1: Are all these steps (lpc, csc, SC, sr and registration) in the correct sequence?

Both the RGB and NB frames should be aligned before splitting.

With regards to the red being "too red", you can adjust the amount of Ha during the HaRGB profile creation in APP, you'll see a slider for Ha.  If my memory serves me correctly, the Bandpass for Ha is 3nm on the L-Enhance, so it is quite narrow, so setting the right blend of Ha during the process will be critical.

I might speak to @mabula-admin on a way to set a defined value based on the NM of the NB filter as I know in other tools, you get to add in this option as well as what multiplier you want to apply to the NB data to the appropriate channel.  I know it all becomes complicated with different filters, but it makes a difference whether the filter is 3nm, 5nm, 7nm, 10nm etc

@michaelacg When you adjust anything, make sure you click on "Re-calculate"

When I add my data, mine gets picked up as Hydrogen Alpha, so what I propose you do is edit the FiTS header of the file and change the Filter to "Ha" and then try again, at the moment, it is assigning it to "Red" instead of "Hydrogen Alpha"

Hi Simon.

Yes, I get the red reduced, by setting the first slider of Ha-Red lower than 2.
I have the 6nm Altair Ha OIII.

And yes, I always click "recalculate".

• But I have problems to get anything meaningful with the Ha channel, created with "Ha-OIII extract Ha".
• If I create the Ha channel by splitting the RGB integration of the HaOIII lights, then I manage to, following your recipe.

@michaelacg How does the "Red" Channel Ha look when using an RGB Split compared to the Red channel on the "Ha-OIII Extract"?

Posted by: @stastro

↑

@michaelacg How does the "Red" Channel Ha look when using an RGB Split compared to the Red channel on the "Ha-OIII Extract"?

See the image in my 15:04 reply --> the one, where I said "a bit too red".

--> Good results with: "Red" Channel Ha look when using an RGB Split, as you told me this morning (your post 11:45)
--> Weird colours, if I use Red channel on the "Ha-OIII Extract"

@michaelacg I meant the Mono images for the Ha Data

Mono Image for Just the Ha Data extracted using RGB Split
Mono Image for Just the Ha Data using Ha-OIII Extract

What do they look like?

 Ah - now I understood your question:

Mono Image = red channel using RGB split of the (normal) RGB integration of the HaO3 lights:

Mono image from Ha-OIII Extract Ha - I kept only the starless version:

And here the Ha-OIII Extract OIII - also starless:

Looking at these 3 images: For my purposes of improving the cigar glow ... the Red-extraction of the RGB looks best.

...

Addendum: Just noticed that I used the red-extract-from-RGB (1st above) with stars.
And the Ha-extract-from-Ha-OIII was used without stars.
--> May this explain the colour behaviour? (If yes - what is the reason?)

Addendum 2: No change, same behaviour (I found the versions with stars). Look here:

@mfc best thing i do with my L-quad and HAO3 filter is: 

1st stack L-quad filter as airy adaptive disc, simple for normal RGB stack.

After that clear everything and load that intergrated L-quad filter and go to tab 2, scroll down and split channels.

Now I get 3 colour channels.

Now the HAO3:  in tab 0 i use the HAO3 extract HA and later O3.

After that is done, clear tab 1 and load R G B HA and O3 intergrated light files.

Do all steps until 5 and after that is done i scroll down to save as normalize lights.

Now I can go to combine RGB, load each channel that you saved as normalize, and play with colors in that tool.

Maybe I need to make a video tutorial 😆 

@mfc best thing i do with my L-quad and HAO3 filter is: 

1st stack L-quad filter as airy adaptive disc, simple for normal RGB stack.

After that clear everything and load that intergrated L-quad filter and go to tab 2, scroll down and split channels.

Now I get 3 colour channels.

Now the HAO3:  in tab 0 i use the HAO3 extract HA and later O3.

After that is done, clear tab 1 and load R G B HA and O3 intergrated light files.

Do all steps until 5 and after that is done i scroll down to save as normalize lights.

Now I can go to combine RGB, load each channel that you saved as normalize, and play with colors in that tool.

Maybe I need to make a video tutorial 😆 

@michaelacg  why you have 2 red? 

Do you use a color camera? So yes.

Than you need to select in tab 0 HA03 extract HA and do the integration and save it as HA.

I guess you used and broadband filter or UV/IR cut? Do the normal process, in tab 0 airy adaptive disc.

After the integration clear tab 1 and load the Broadband intergrated light file. Go to tab 2 and scroll down, select split channel and save that.

Now you got 4 channels, R G B HA. Clear tab 1 and load each color and select what they are, APP always ask for what color filter it is.

Do step 2 until 5, when 5 is done, scroll down and save as normalize.

Now you see 4 new files down below, go to tools and do combine RGB, load the 4 normalize channels and play with colors.

Hi Stefan.

My system is
- OSC camera (ASi2600MC)
- 6nm Ha OIII filter

I did the following sessions and integrations:

1. 3 session, total 7.7 hours, just OSC
   --> integrated with adaptive airy disc, Bayer/X-trans drizzle 1.5 with droplet 1.5, rest standard
   --> split channel, getting R, G, B
2. 1 session, 3 hours with the Ha OIII filter
   --> a. integrated as RGB, same parameters as above --> split channel, getting filterR, filterG, filter B
   --> b. integrated with (in tab 0) "Ha-OIII extract Ha" and "Ha-OIII extract OIII"

I used Simons advice:

Posted by: @stastro

↑

Re-Construct the image with the R (Ha), R, G, B using the RGB tool using the HaRGB profile

... following this, you get 2 red (Ha, R) and 1 G and 1 B.

• If I use filterR from above as Ha --> I get good colours, and can play with the Ha-red to tune the intensity of red.
  .
• If I user Ha-OIII extract Ha as Ha --> all my tries result in these weird colours. 
  I must do something stupid, because both are mono channels and should result in comparable colours.

-----

Posted by: @stefanastro

↑

@mfc best thing i do with my L-quad and HAO3 filter is: 

1st stack L-quad filter as airy adaptive disc, simple for normal RGB stack.

After that clear everything and load that intergrated L-quad filter and go to tab 2, scroll down and split channels.

Now I get 3 colour channels.

Now the HAO3:  in tab 0 i use the HAO3 extract HA and later O3.

After that is done, clear tab 1 and load R G B HA and O3 intergrated light files.

Do all steps until 5 and after that is done i scroll down to save as normalize lights.

======================

Now I can go to combine RGB, load each channel that you saved as normalize, and play with colors in that tool.

Maybe I need to make a video tutorial 😆 

• Up to the ==== line, I understand. And (think) have followed.
• Can you elaborate, how you do "combine RGB". Yes, video would be good. Or step by step.

@michaelacg Yes, your Ha Extract must include stars to be merged into the HaRGB image, stars need to be present for sure.

Looking at the screenshots, it looks like the OIII Extract has similar detail to the Ha Extract from the RGB Split, so as a test, use the OIII Extract and merge that into the RGB using the HaRGB profile and see what the resulting image is

@stastro 

Looks perfect, very similar to R-extract-from-RGB.

Here the screenshot from HaRGB with O3-extract-from-Ha-OIII and split R, G, B channels.

==> So the learning: Combine RGB need the stars.
==> Might be added as a hint in the menu.

But - doing the same with the Ha-extract-from-Ha-OIII:

... strange.

And it was pure luck that I used the R-from-RGB as pseudo-Ha with stars. Simply forget to do "sr".

Hi Stefan, I assume your response was directed to Michael and not myself? Irrespective, it brought my attention back to this thread and I see there are a few new tips that may help me. I now have an Ha / O3 L-Para filter as that allows me to use it at both f2.2 and f10 on my SCT. So processing with it should be similar to with your quad-band filter.
A video with the latest suggestions would be great. 

From my perspective there’s no rush as, at 61N, I won’t have dark skies again until September… 😖 I now have 5 months of pure lunar and solar astrophotography ahead of me.