Monday, 16 November 2026

How to process a multi-night dataset in SIRIL

Want to find something quickly on my blog? Use this page: https://undersouthwestskies.blogspot.com/p/how-to-find-topicpost-quickly-on-my-blog.html


M31 Andromeda Galaxy: My First Multi-Night Dataset in Siril

I've photographed M31, the Andromeda Galaxy, several times over the last few years. Every attempt has taught me something new, usually after several hours of confusion, a handful of forum searches, and the odd moment of staring blankly at the screen wondering what on earth I've just done.

This latest image marks an important milestone for me because it is the first time I have successfully processed a multi-night dataset in Siril.

The image below combines data gathered across three separate nights and, while there are undoubtedly quicker and more elegant ways to do this, I thought it would be useful to share exactly what I did in case it helps someone else who is currently wrestling with the same learning curve.

After all, that's what this blog is about: sharing successes, sharing mistakes, and hopefully helping all of us move forward a little faster.

 


M31 Andromeda Galaxy Capture Details

Date

12th November 2026

Location

Wembury Beach Car Park, South Devon

Observing Conditions

The weather was about as good as I could have hoped for:

  • Perfectly clear skies
  • Good seeing conditions
  • Bortle 4 skies
  • Temperatures just below 0°C

It was one of those sessions where the sky seemed determined to cooperate while the temperature tried its best to freeze the astrophotographer operating underneath it.

In fact, by the end of the session I had to scrape ice from the car windscreen before driving home.

 

Equipment Used

  • Zenithstar 61 II with 61A field flattener
  • Astromodded Canon 800D
  • Sky-Watcher EQM-35 Pro Mount
  • ASIAIR Mini
  • ZWO 120 Mini guide camera
  • RVO 32mm guide scope
  • Celestron Lithium Pro PowerTank
  • Three separate power banks for camera and dew control

 

Capture Settings

Across three separate nights I captured the following on each session:

Light Frames

  • 20 × 30-second exposures
  • 20 × 120-second exposures
  • 20 × 180-second exposures

Calibration Frames

  • 20 darks
  • 20 bias frames
  • 40 flats

Processing Software

  • Siril
  • Affinity Photo 1

 

Processing a Multi-Night M31 Dataset in Siril

This was where the real challenge began.

Until now, I had only processed single-night datasets. Combining multiple nights containing several different exposure lengths felt a bit like being handed three separate jigsaw puzzles, mixing all the pieces together and then being told there was supposedly only one final picture.

I searched forums and watched videos, but somehow managed to become more confused rather than less.

So eventually I did what many astrophotographers do when faced with uncertainty:

I made up my own workflow and hoped for the best.

Fortunately, it actually worked.

 

Stage 1: Create Master Lights for Each Night

For each individual night, and for each exposure length, I ran Siril's:

OSC_Preprocessing.ssf script

This calibrated, registered and stacked each set of images independently.

The result was a collection of calibrated and stacked master light files for:

  • Night 1
  • Night 2
  • Night 3

at

  • 30 seconds
  • 120 seconds
  • 180 seconds

In other words, I ended up with multiple completed master stacks ready for combination.

 

Stage 2: Create a Master Lights Folder

Next, I copied all those stacked master light FITS files into a brand new folder.

This became my working directory inside Siril.

Think of it as gathering all the ingredients onto the kitchen counter before attempting the recipe.

 

Stage 3: Build a Sequence

On the Conversion tab I:

  • Added all the master light files
  • Created a sequence called M31_Masters
  • Enabled Symbolic Links
  • Clicked Convert

Using symbolic links meant Siril referenced the original files rather than duplicating them, saving both time and disk space.

 

Stage 4: Register the Master Stacks

Before registration, I selected my cleanest 180-second master stack as the reference image.

(Exactly where I did this in the workflow is unfortunately lost somewhere between memory and old age. If I find the precise step again, I'll update this post!)

I then moved to the Registration tab and selected:

Global Star Alignment

before clicking:

Go Register

This aligned all the master stacks onto a common reference frame.

 

Stage 5: Stack the Registered Masters

On the Stacking tab I used the following settings:

Stacking Method

Average stacking with rejection

Normalisation

Additive with scaling

Output Normalisation

Enabled

RGB Equalisation

Enabled

Pixel Rejection

None

Weighting

Noise or Number of Frames

Then I clicked:

Start Stacking

 

Why I Chose These Settings

One thing I've learned over the years is that button-clicking without understanding is a dangerous game in astrophotography.

Sometimes it works.

Sometimes it produces magic.

Sometimes it produces modern art.

Here's the reasoning behind my choices.

 

Additive With Scaling

My 30-second, 120-second and 180-second master stacks all contained different brightness levels and background sky values.

Without normalisation, trying to combine them would be like asking three musicians to perform together while each played in a different key.

Additive with scaling aligns the background levels and scales brightness values so the stacks can work together without clipping bright regions such as M31's core.

 

Noise Weighting

This was the setting I was most interested in.

The idea is beautifully simple.

Longer exposures usually contain more useful signal, particularly in the faint outer dust lanes and halo structures.

Noise weighting evaluates the signal-to-noise ratio of each master stack and automatically gives greater importance to the cleaner data.

In theory, the deep 180-second stacks contribute more to the faint galaxy structure, while the shorter exposures help preserve cleaner bright-core detail.

At least that was the plan.

Thankfully, it seemed to work.

 

Output Normalisation

I enabled output normalisation to ensure the final combined image remained safely within Siril's floating-point data range.

My thinking was that I wanted to preserve every last scrap of dynamic range before applying a Veralux Hyperbolic Stretch later in the workflow.

 

Why I Disabled Pixel Rejection

By this stage I only had a small number of master files.

The original calibration and stacking runs had already dealt with:

  • Hot pixels
  • Satellite trails
  • Random artefacts

Pixel rejection algorithms generally perform best when larger numbers of frames are available.

I was concerned that running rejection again might remove legitimate data from M31 rather than noise.

So, I left it switched off.

 

The Result

Rather surprisingly...

It worked.

The final stack produced a clean linear master file ready for post-processing.

The very first thing I did was rename result.fit before accidentally overwriting it.

Experience has taught me that unnamed processing files have a life expectancy roughly equivalent to a chocolate teapot.

 

My Siril Processing Workflow

Once the master linear image was complete, this was my processing sequence:

  1. Open result.fit
  2. Enable AutoStretch display mode
  3. Rotate and crop
  4. Run Plate Solver
  5. Veralux Nox
  6. SPCC
  7. Cosmic Clarity Sharpening
  8. Silentium
  9. StarNet
  10. Return display mode to Linear
  11. Veralux Hyperbolic Stretch
  12. Veralux Curves
  13. Contrast Limited Adaptive Histogram Equalisation (CLAHE)
  14. Veralux Revela
  15. Save
  16. Veralux StarComposer
  17. HDR Multiscale
  18. Veralux Vectra
  19. Final Save

I will write a separate post explaining the thinking behind the order of these stages because there is quite a bit of reasoning behind it.

 

Lessons Learned

The biggest lesson from this project was that multi-night datasets aren't nearly as frightening as they first appear.

In reality, the process is simply:

  1. Calibrate and stack each night's data correctly.
  2. Treat those stacks as master subframes.
  3. Register them together.
  4. Stack them again.

Once I started thinking of each night's result as just another "light frame", the workflow suddenly made much more sense.

Like many things in astrophotography, the hardest part is often convincing yourself to start.

 

M31 Then and Now

One of the joys of astrophotography is seeing progress over time.

Below are my previous images of M31, shown in the order they were captured.

Looking back at old images can sometimes be uncomfortable, but it is also one of the best ways to see how far you've come.

Every image represents another lesson learned, another mistake overcome, and another step forward.

These were the first two images I did of M31 using my Samyang 135 lens on my Canon 800D before it was astromodded. The mount was a SWSA2i tracker and there was no guiding 


This was done with the Zenithstar on the SWSA2i - yep I did do that! 


And back to my latest one. I'm quite proud of it - it shows some substantial progress at long last


Final Thoughts

There's still plenty for me to learn and I'm sure there are more efficient workflows than this one.

If you process multi-night datasets differently, I'd genuinely love to hear about it.

Drop a comment below and share your approach. The astrophotography community is at its best when we're all comparing notes, sharing ideas and helping each other avoid the mistakes we've already made ourselves.

Until next time, stay safe, good viewing, and keep looking up.

Steve
PlymouthAstroBoy




Want to find something quickly on my blog? Use this page: https://undersouthwestskies.blogspot.com/p/how-to-find-topicpost-quickly-on-my-blog.html

No comments:

Post a Comment

Hello, thanks for visiting us. If you want to drop a comment or get in touch - leave a comment and we will get back to you asap. Bye