Begging some advice from mosaic and Pixinsight gurus

[old_eyes]

I have just completed a three pane mosaic of the regions from the Clamshell to the Pelican nebulas. Most of my problems were struggling to balance up the three panels to avoid visible joins. But whilst pixel-peeping to see if I was having any success (always a dangerous game) I noticed some misalignment issues.

This is the edge of the frame where a couple of panels overlap:

Clearly poor alignment and/or distortion.

Take an image at the same scale and column position, but from teh centre of the frame and you get this:

Not the same problem at all.

I am looking for advice on what I might have done wrong, how I might fix it, or whether I should just be grateful that our modern capabilities tempt me into such an insane project.

This was an Ha, O3, S2 image in SHO palette. Processing in Pixinsight

My process was:

1. Get all the 9 frames calibrated and stacked
2. DynamicBackgroundExtraction on each frame
3. Combine the frames for each filter into a mono mosaic:
   1. DNA LInear Fit two outer panels for each filter to the centre panel (strongest signal)
   2. Image solve each panel
   3. Run MosaicByCoordinates for the three panels using script defaults
   4. GradientMergeMosaic using Overlay combination method with increased shrink and feather radius to suppress seams.
4. Nonlinear stretch each mono mosaic and balance the intensities (as well as I could)
5. Register S2 and O3 mosaics to Ha using StarAlignment with Projective Transformation mode and Register/Match Images mode - Frame Adaptation Checked.
6. ChannelCombination to convert to RGB mosaic.

I wonder whether I would have been better to non-linear stretch the frames before creating the mosaics.

Could the problem be that I needed to use some of the Mosaic Geometry options in MosaicByCoordinates? To be honest I would not know where to start.

Any advice gratefully received, including - "nobody will notice"!

[tomato]

Have you had a try with the Photometric Mosaic script in Pixinsight? It worked wonders on the data I collected for a 12 panel LRGB mosaic of M31 last year. You will need a wide field image of the region to act as a reference image.

I am no expert with the script but with help from  @Laurin Dave, the basic workflow I followed was:

“the first step will be to create the individual panels as stand alone items...  next crop these so that there are no artefacts at the edges..  this is important as any artefacts will mess up the gradient removal..   next step is to plate solve each panel using Pixinsight's Script... Image Analysis..   Image Solver.... once this is done you need to co-register them all, this is done using Pixinsight Script... Utilities.... Mosaic by Co-ordinates..... now would be the time to a) go for a long walk or do overnight...  

Once you have all the registered panels, including the wide field reference frames, the fun starts with the Mosaic script...”

The script help file is very useful.https://drive.google.com/file/d/1sxPP-L2WMMQEESTsZVFEtQr8xr1ugti4/view
I note your data is NB but this shouldn’t be a problem.

[Laurin Dave]

Did you use “fast” or “high quality” as the option in Mosaic by Coordinates? If the former then try the later..  and do it overnight as it’ll take a while   also do the Ha, Oiii and Sii for each individual panel align correctly?  If not then I suggest you register the subs using thin plate splines and distortion correction ..  

[old_eyes]

12 hours ago, Laurin Dave said:

Did you use “fast” or “high quality” as the option in Mosaic by Coordinates? If the former then try the later..  and do it overnight as it’ll take a while   also do the Ha, Oiii and Sii for each individual panel align correctly?  If not then I suggest you register the subs using thin plate splines and distortion correction ..  

Thanks, I will check those options out. Aligning panel subs to each other before mosaic aligning makes sense. 

[old_eyes]

On 20/08/2021 at 11:18, old_eyes said:

Thanks, I will check those options out. Aligning panel subs to each other before mosaic aligning makes sense. 

So.

I aligned the alls the subs for the various panels to the Ha filter and chose "high quality" in Mosaic by Coordinates.

That did not improve things significantly as can be seen from the Ha mosaic at the edge:

You can see the same problems in the boundary area.

I next tried the Photometric Mosaic script and it seems to work well. No trace of the misalignment and fewer problems with stars on the panel boundaries distorting and leaving artifacts:

I don't understand how it works compared to the Gradient Mosaic Merge, but it seems to be a success.

Thank you @tomato and @Laurin Dave

 

[old_eyes]

This is the new raw mosaic after ChannelCombination:

Nothing has been done to the image except Invert-SCNR-Invert to blitz purple star haloes, so plenty more to gain I hope! Edges look pretty good.