Stacking two highly overlapping panels together causes some mis-registered stars

I’m sometimes stack “mini mosaics” with two panels with a high level of overlap to slightly increase my field of view. I recently did this for the Rosette nebula with two panels with 65% overlap. Then I can just stack all the lights together and this often works great with SIRIL. It works better than stacking each panel and then stacking the results together because there are no seams between the panels. It’s silky smooth.

But every once in a while, there are areas of the image where they don’t properly register and the stars are misaligned. This time it happened with the stars shifting out of alignment on a portion of the right side of the image.

Just for grins, I tested in APP with defaults, and it has the same issue so I certainly don’t consider it a SIRIL bug, just I threw SIRIL a complex case. But in APP, if I set “dynamic distortion correction” which corrects for optical distortions between frames, it stacks correctly.

I stack with the OSC processing scripts, so it uses a lot of default options. Is there a registration option I can set that will give SIRIL additional freedom in registering to correct for, what I assume is, optical distortion that exceeds SIRILs normal parameters between my highly overlapping panels?

I can try to find a small set (2-3) lights that do it if needed.


In the current version, distortion is not taken into account. This is something that will be handled in the 1.4, not before.

1 Like

Excellent that it’s planned for the future! It’s already a top notch stacker and so fast!

Hi @smiller ,

would be interested to get your images indeed!



1 Like

I would be happy to share, although I do take very short exposures as I shoot with a stock unguided Alt/Az Goto Dobsonian (Orion XT12G Goto 12" Dob), so probably too many lights to share the whole night’s capture.

But thanks to low read noise cameras, fast computers and a fast but high quality stacker (SIRIL!! Yay!) I can stack several thousand subs pretty darn quickly. I’m very pleased with my results given I had no expectations.

I would certainly be happy to share data, but I have so many exposures, although I could just share a few of each panel as it only takes a few to check registration. So if you want a sample of those, let me know.

For completeness here is my astrobin of shots from this Dobsonian, almost all stacked with SIRIL. The latest image, the Rosette Nebula was the latest example of a “mini mosaic” I was referring to and it would be lights from that image I would share:

Yes, that’s exactly what I would need. Just a few of them, otherwise, my computer will just explode :slight_smile:


The joys of Siril development, when you realize that your computer has more images of other people than your own. ^^

Here is a link to a small set of the files:

  1. 24 Lights, 12 for Panel 1 and 12 for Panel 2 (you can tell by the “P1” or “P2” in the name). They were captured in NINA with RICE compression which is why they are 1/2 the expected size.

  2. A small subset of Biases and Flats to use.

  3. The script I used.

To stack I use two pass registration with the -framing=max option. For your convenience I attached the script I use to stack that is one of the standard OSC preprocessing scripts modified for no Darks, to use two pass registration with -framing=max and to use 16-bit precision for calibration and registration frames but 32-bit precision for the final stacking. This saves disk space.

Note, these were captured with very short exposures (8 seconds) as I use an unguided Alt/Az Goto Dob meant only for observation. Also, they have very bad vignetting on the very ends because I use a full frame ASI2400MC along with the Nexus 0.75X reducer which is only designed for APS-C sized sensors, so it creates a significant vignetting shadow on the edges. I often trim the width of the resulting stack 10-15% as a result.

In any case, the misregistration happens to the right and left of center and gets worse as you get towards the edges, especially on the right side.

And just for completeness, the final processed image is here: (Rosette Nebula (NGC 2246) (Steven Miller) - AstroBin)

Hi steven,

thx for the images. Just to make sure I understand fully your workflow, do you:

  • apply your script on P1 and then P2 lights and then register P1 and P2 stacks together
  • or apply your script on all P1/P2 images at once?

I’m saving these for later anyway, I’m finishing smthg else at the moment. I’ve just opened one image and solved it with distorsions to understand what could cause your problem. The reducer does not only create the vignetting but also some severe optical distorsions, which are the cause of the misregistration:

Hopefully, this will be handled in next version!
Thanks again for the share, will keep you posted.




In this case because of the large amount of overlap, which was planned, I intended to stack them all together.

However, with true mosaics with less overlap, I will stack each panel independently and then stack the panels together, usually as a mosaic.

Wow, that does appear to be a lot of optical distortion! Are those lines really representative of the magnitude or are they exaggerated for emphasis. What tool is that by the way? (I now realize I need to learn to measure mine)

I know the Nexus corrects up to APS-C only so I should expect some but not much until the extreme edges. Perhaps I have some backspacing to play with.

Yes they are.

This is done with Siril dev version.

1 Like

There’s exaggeration indeed, still need to work on that tool. In particular, i want to give a measure of the worst distorsion. In your case, it’s in the order of 50" in the corners. As mentionned by @lock042, it’s in Siril dev version. Not sure there’s any BF problem, because the center seems very flat. Sorry it came out that scary…

You guys get to play with all the cool tools! :wink:

Thanks for the info on my distortion, that what certainly over and above the call of duty.

I need a good scare now and then…

Let me know if there is any other help or info I can provide.



Just giving some heads up on the distortion/stacking dev…
Here is the final stack with the stuff going on with astrometric registration and undistortion…


And a close-up view in the overlapping region on the right

Still need to do a lot of clean-up but I may ask you to test in a foreseeable future :slight_smile:



1 Like

Wow, this is an amazing advancement! I’m starting to think Christmas will come early this year!

1 Like