Back to forums

Help Needed with Stacking Multi Session Narrowband Data in Siril

Siril Team Siril Narrowband Duo-Narrowband NGC6888 NGC Objects
18 replies•941 views
Dominik avatar
Dominik avatar
Hello,

i'm a newbie, and I would like to process Narrowband data using Siril for the first time. I need some advice.

I have about 24 hours of raw data from NGC 6888, captured with an ASI 533MC camera and a Duo Narrowband Filter (Askar Colour Magic 6nm HA/OIII).

As a test, I tried stacking a portion of the imaging sessions using Sirilic, DSS, and directly with Siril (RGB). However, in Siril, I cannot extract the HA and OIII data using the CFA Split option (CFA Split is greyed out and cannot be selected) to further process them and later combine them using Pixel Math.

Does the CFA Split only work with mono captures? Am I doing something wrong during stacking? As far as I have read, the CFA data should be in the FITS file header.

As an experiment, I took a single session and stacked it using the HA/OIII extraction script, and it worked perfectly.By the way, is this method even correct? I would be grateful for any tips.

I also noticed when using this HA/OIII extraction script for a single session that the OIII data is significantly larger than the HA data. Is that correct? Could this be due to the Bayer matrix (Green + Blue)?

Clear skies

Dominik
Respectful Engaging
andrea tasselli avatar
andrea tasselli avatar
I'm afraid you need to post the Siril log file to try to understand what you did (and didn't do). But I can offer an answer for the last question, which is straight from the Siril help file:=commandnameextract_HaOIII =highlight-text notranslate=highlight
extract_HaOIII [-resample=]
=line-block=lineExtracts Ha and OIII signals from a CFA image. The output files names start with the prefix "Ha_" and "OIII_"=line =lineThe optional argument -resample={ha|oiii} sets whether to upsample the Ha image or downsample the OIII image. If this argument is not provided, no resampling will be carried out and the OIII image will have twice the height and width of the Ha image=line-block=line

The most strightforward way is to calibrate the images as stright FITS files and then Debayer (not CFA split) them. Now the OIII signal is contained in the G and B channels and the Ha in the R channel. If you process them as if they were "normal" color channels the results is guranteed to be pleasing enough assuming standard color calibration at the end. Some people elect to drop the B channel (as it is the noisiest in most instances) and use just the G and R channel, duplicating the G channel for the B channel and thus producing an OOH/HOO image.
Helpful
Dominik avatar
Dominik avatar
The most strightforward way is to calibrate the images as stright FITS files and then Debayer (not CFA split) them. Now the OIII signal is contained in the G and B channels and the Ha in the R channel. If you process them as if they were "normal" color channels the results is guranteed to be pleasing enough assuming standard color calibration at the end. Some people elect to drop the B channel (as it is the noisiest in most instances) and use just the G and R channel, duplicating the G channel for the B channel and thus producing an OOH/HOO image.

Thank's for your help!

I'm trying to understand how to process DUO Narrowband data with Siril and what the steps look like.

I used the Built-In script OSC_Extract_HaOIII.ssf for my test sample (single session). In this script, additional arguments cannot be specified or can I do it after all?

If I understood correctly, I just need to calibrate and debayer the FITS files in Siril itself. Do I need to do this for each of the sessions?

I have a total of 10 individual sessions, each with flats, darks, dark flats, and bias files. I thought it would be better to handle all of this in one step using Sirilic.

In relation to NGC 6888, the blue filaments are indeed of particular interest, and I want to emphasize them. That's why I separated them for later processing in the HOO method.

In this case, would it be meaningful to remove the B (Blue) channel and duplicate the G (Green) channel?

Clear skies
Dominik
Sergey Volkov avatar
Sergey Volkov avatar
andrea tasselli:
I'm afraid you need to post the Siril log file to try to understand what you did (and didn't do). But I can offer an answer for the last question, which is straight from the Siril help file:=commandnameextract_HaOIII =highlight-text notranslate=highlight
extract_HaOIII [-resample=]
=line-block=lineExtracts Ha and OIII signals from a CFA image. The output files names start with the prefix "Ha_" and "OIII_"=line =lineThe optional argument -resample={ha|oiii} sets whether to upsample the Ha image or downsample the OIII image. If this argument is not provided, no resampling will be carried out and the OIII image will have twice the height and width of the Ha image=line-block=line

The most strightforward way is to calibrate the images as stright FITS files and then Debayer (not CFA split) them. Now the OIII signal is contained in the G and B channels and the Ha in the R channel. If you process them as if they were "normal" color channels the results is guranteed to be pleasing enough assuming standard color calibration at the end. Some people elect to drop the B channel (as it is the noisiest in most instances) and use just the G and R channel, duplicating the G channel for the B channel and thus producing an OOH/HOO image.

*** I followed the script in this video, though not quite step by step, but I followed the basic principle, try it, maybe it will improve your result https://www.youtube.com/watch?v=fDhOrKvM7GU***
Dominik avatar
Dominik avatar
Sergey Volkov:
=14px*** I followed the script in this video, though not quite step by step, but I followed the basic principle, try it, maybe it will improve your result https://www.youtube.com/watch?v=fDhOrKvM7GU=14px***

I know the video. Unfortunately, it only demonstrates a single session and does not show the processing of multi-sessions.
Well Written
Sergey Volkov avatar
Sergey Volkov avatar
Dominik:
Sergey Volkov:
=14px*** I followed the script in this video, though not quite step by step, but I followed the basic principle, try it, maybe it will improve your result https://www.youtube.com/watch?v=fDhOrKvM7GU=14px***

I know the video. Unfortunately, it only demonstrates a single session and does not show the processing of multi-sessions.

*** I also handled multisession. I just collected all the lights in one folder, the calibration frames were common. including flats. I did not rotate the camera for three sessions and everything turned out with the division into hydrogen and oxygen according to the script. ***
Joe Linington avatar
Joe Linington avatar
I will do my best here, when processing OSC dual narrowband data, you can just debayer it and then stack it. This will give you an HOO image. If you want to separate out the channels (which can lead to some quality improvement) then you can use the Extract_HAOIII (Seqextract_HAOIII) to pull the specific data from un-debayered (CFA) images. In other words you have to ensure you omit the -debayer option when you calibrate. When you stack extracted Ha/Oiii data you need to drizzle some of it back to the proper size. the Ha/Oiii script has all of those commands in it.

For multiple sessions there 2 ways I tackle this problem. The first way is to have scripts that only calibrate your images (ready for registration) and use the -start=(an index number here) with the convert command. Each night starts with an index number high enough that each filename is unique. I usually use 300 for night 2, 600 for night 3 etc. I can then move all of these calibrated frames into a single folder and register and stack them either using a second script or manually. The other handy command for multi nights is the Merge command which lets you merge a bunch of sequences into a single sequence. The command handles the numbering so no need to use "convert -start=index" if you are going to use merge.

A lot of this is done in custom scripts. This is where Siril gets a little unix/linux like in it's use. There are some very powerful options but it's a little bit to get used to if you don't come from that world. Astro Pixel Processor is where most people turn when Siril gets a bit silly. It is quirky in it's own way but is a fantastic stacking program that has some very powerful features and I find it works very well with the rest of the astrophotography software ecosystem, both open source and as a fantastic alternative to WPBB  in pixinsight.
Helpful
andrea tasselli avatar
andrea tasselli avatar
Dominik:
I'm trying to understand how to process DUO Narrowband data with Siril and what the steps look like.

I used the Built-In script OSC_Extract_HaOIII.ssf for my test sample (single session). In this script, additional arguments cannot be specified or can I do it after all?

If I understood correctly, I just need to calibrate and debayer the FITS files in Siril itself. Do I need to do this for each of the sessions?

I have a total of 10 individual sessions, each with flats, darks, dark flats, and bias files. I thought it would be better to handle all of this in one step using Sirilic.

In relation to NGC 6888, the blue filaments are indeed of particular interest, and I want to emphasize them. That's why I separated them for later processing in the HOO method.

In this case, would it be meaningful to remove the B (Blue) channel and duplicate the G (Green) channel?

Clear skies
Dominik


In Siril I don't use scripts so it is hard for me to say. But as I and Joe said your best way around is to process these files like ordinary CFA (RGB) files and do whatever manipulation afterwards. You need to calibrate each and every session on its own, since flats at the very least can be different (but many other things too). The OIII filaments in NGC6888 are obviously well preserved in the G channel as well as, to a lesser extent, the B channel. If you were to follow the HOO path (thus duplicating the G channel to the B channel) it makes sense do remove the B channel. If, on the other hand, you were to create a RGB image you're well advised to keep the B channel in its place, IMO. In my view is best to process each single session (night) on its own but then I use PI so that I can use optimization strategies and tools on multi-session compositions. Not sure you can do the same in Siril but you won't be none the worse if you were to follow the same path.
Doug Crowe avatar
Doug Crowe avatar
I used Sirilic for the first time last week on 5 nights of ngc7000 with my L-extreme. I followed this video  Stacking Multiple Night Sessions using Siril and Sirilic - YouTube by Deep Space Astro. I ran it twice getting a normal RGB first and then separating it into Ha and Oiii. The Oiii is a larger image but I use pixinsight too and used star alignment to make them the same size. I haven't had time to process the Ha and Oiii to see how they will come out.
daywalker avatar
daywalker avatar
Dominik:
Hello,

i'm a newbie, and I would like to process Narrowband data using Siril for the first time. I need some advice.

I have about 24 hours of raw data from NGC 6888, captured with an ASI 533MC camera and a Duo Narrowband Filter (Askar Colour Magic 6nm HA/OIII).

As a test, I tried stacking a portion of the imaging sessions using Sirilic, DSS, and directly with Siril (RGB). However, in Siril, I cannot extract the HA and OIII data using the CFA Split option (CFA Split is greyed out and cannot be selected) to further process them and later combine them using Pixel Math.

Does the CFA Split only work with mono captures? Am I doing something wrong during stacking? As far as I have read, the CFA data should be in the FITS file header.

As an experiment, I took a single session and stacked it using the HA/OIII extraction script, and it worked perfectly.By the way, is this method even correct? I would be grateful for any tips.

I also noticed when using this HA/OIII extraction script for a single session that the OIII data is significantly larger than the HA data. Is that correct? Could this be due to the Bayer matrix (Green + Blue)?

Clear skies

Dominik

did you select the debayer before saving option on the calibration tab when stacking with siril?
Dominik avatar
Dominik avatar
=14pxdid you select the debayer before saving option on the calibration tab when stacking with siril?

No, in the sample with just a single session, I did not choose that option. In Sirilic, there didn't seem to be such an option at first glance. I might manually calibrate and debayer each individual session now.
daywalker avatar
daywalker avatar
thats what i do
Dominik:
=14pxdid you select the debayer before saving option on the calibration tab when stacking with siril?

No, in the sample with just a single session, I did not choose that option. In Sirilic, there didn't seem to be such an option at first glance. I might manually calibrate and debayer each individual session now.

thats what i do.. well sort of.Via a script.
Aaron H. avatar
Aaron H. avatar
Dominik:
However, in Siril, I cannot extract the HA and OIII data using the CFA Split option (CFA Split is greyed out and cannot be selected) to further process them and later combine them using Pixel Math.

It sounds like your images may already be debayered. The "Split CFA" option only works with non-debayered CFA images. It will split each image into four images with half the vertical and horizontal resolution (according to the Bayer pattern).

If your image is displaying in colour in Siril, then it's already debayered and Split CFA cannot be used.

Personally, my process for dual-narrowband data with Siril is to stack normally as if it was a broadband image, then to separate the stacked image using "Split Channels". Red then becomes Ha and the G and B are combined using PixelMath for OIII (for my camera and filter 0.75*G + 0.25*B works well).
Well Written Helpful Insightful
Dominik avatar
Dominik avatar
Aaron H.:
Dominik:
However, in Siril, I cannot extract the HA and OIII data using the CFA Split option (CFA Split is greyed out and cannot be selected) to further process them and later combine them using Pixel Math.

It sounds like your images may already be debayered. The "Split CFA" option only works with non-debayered CFA images. It will split each image into four images with half the vertical and horizontal resolution (according to the Bayer pattern).

If your image is displaying in colour in Siril, then it's already debayered and Split CFA cannot be used.

Personally, my process for dual-narrowband data with Siril is to stack normally as if it was a broadband image, then to separate the stacked image using "Split Channels". Red then becomes Ha and the G and B are combined using PixelMath for OIII (for my camera and filter 0.75*G + 0.25*B works well).

Yes, it is debayered. Sirilic runs completely from calibration to stacking. I will try your method and see what happened. I am also testing other methods to see how well it works.
Dominik avatar
Dominik avatar
Thank you for all the tips and suggestions. I think I'm going to be busy for a while now smile


I think I'll create a small sample size with data for the tests. Makes a lot easier.

CS
Dominik
s_pls avatar
s_pls avatar
Doug Crowe:
ran it twice getting a normal RGB first and then separating it into Ha and Oiii

I have been trying to run the Ha Oiii extraction on multiple nights for a while with no success! I'm also a mega beginner so apologies but can you run me through this process? I have watched the video from Deep space Astro, but I don't get how to go from RGB to having the separate channels? Thanks for the help!
Engaging
Dominik avatar
Dominik avatar
Doug Crowe:
ran it twice getting a normal RGB first and then separating it into Ha and Oiii

I have been trying to run the Ha Oiii extraction on multiple nights for a while with no success! I'm also a mega beginner so apologies but can you run me through this process? I have watched the video from Deep space Astro, but I don't get how to go from RGB to having the separate channels? Thanks for the help!

Why don't you try it directly with Sirilic? Sirilic is a graphical add-on for Siril that helps you with stacking.

With it, you can stack different nights and let the result of a DNB capture be output directly as OIII and Ha (Greyscale)

After that, you can generate an individually weighted color image from OIII and HA using Pixelmath.

Sirilic YT

CS 
Dominik
s_pls avatar
s_pls avatar
Dominik:
Doug Crowe:
ran it twice getting a normal RGB first and then separating it into Ha and Oiii

I have been trying to run the Ha Oiii extraction on multiple nights for a while with no success! I'm also a mega beginner so apologies but can you run me through this process? I have watched the video from Deep space Astro, but I don't get how to go from RGB to having the separate channels? Thanks for the help!

Why don't you try it directly with Sirilic? Sirilic is a graphical add-on for Siril that helps you with stacking.

With it, you can stack different nights and let the result of a DNB capture be output directly as OIII and Ha (Greyscale)

After that, you can generate an individually weighted color image from OIII and HA using Pixelmath.

Sirilic YT

CS 
Dominik

I have tried but I have no idea how to use that script really. So do I need to edit a script/copy the HA Oiii script and paste in Sirilic? Thanks
Dominik avatar
Dominik avatar
I have tried but I have no idea how to use that script really. So do I need to edit a script/copy the HA Oiii script and paste in Sirilic? Thanks

You don't need to edit or insert anything in Sirilic. The Sirilic graphical add-on as an interface has parameters that need to be clicked. The script then runs in the background and the finished result ends up in the 'Work' directory. It's explained in the video.

CS 
Dominik