Mosaic Star Alignment Problem after 10/12

Jastar

Active member
Hey Guys since 2018 I have data on my harddisc from Top of the Mountain Roque de los Muchachos on La Palma where I shot a 12 Panel Mosaic of the Milkyway with a 50mm lens.

When I wanted to process it, I became error messages so I didnt process them back then.
Now 3 years later I like to give it another try.
I Already figured it out to to register 10 pieces together, but for the last 2 I always get an error message
Error: Unable to find an initial set of putative star pair matches.

I tried different settings for the whole afternoon now, I attache the setting for which it worked for the other 10 panels.

It always stucks at "computing intersections" in the process explorer.

Please could somebody give me an advice how I can complete this project?
I would very appreciate it
PI version: 1.8.8.7

thanks in advance Simon

Sorry for the double Post could a Moderator please delete the first one? Seems like I can't delete
 

Attachments

  • mosaik.jpg
    mosaik.jpg
    185.3 KB · Views: 54
Last edited:
i think probably solving each pane, then using the script called mosaicbycoordinates and finally the script called photometricmosaic is probably the right way to go here. with staralignment it gets progressively harder to register the panes.

as for this technique, are you using previews to restrict the star matching?

rob
 
Hi Rob Oh gosh something completely new 1 m before the finish line ;-P

Meanwhile I managed to get No. 11 together so actually still fighting with no 12.
For 11 I changed following settings:
Ransac Tolerance = 8
noise Reduction = 10
Maximum Distorttion = 700

No I don't work with previews, so I currently anyway first struggling with the "Master reference Image" to later get into the smooth combining process.
 
you can try defining previews around the areas where you need to join the panes (on the mosaic and on the last pane that won't register, on the parts that should overlap) and that might succeed.

rob
 
If you are down to two frames, and you can solve plate solve them (and the other 10 joined), then you could used the script AlignByCoordinates to finish the job (or at least the alignment; the background matching is another matter...).
 
Ok Guys just took my 11 Panel image and processed it further with "register match images" and then "Gradient merge mosaic".

Look what silly result i got... looks like modern art ;-P
And the second what it should be, left bottom would be no 12 which could never be added...
Edit added a third pic, this is the result with more but not exactly "standard values" during the "register match images" and then "Gradient merge mosaic" process, but also a very poor result on the areas where frames overlap -> not usable.

Seriously is there a working way to create a 12 panel mosaic in Pixinsight?
Please im no Pix Pro, I need a detailes tutorial (best video) which I could follow. :-(
 

Attachments

  • art.jpg
    art.jpg
    476.2 KB · Views: 60
  • Unbenannt-4.jpg
    Unbenannt-4.jpg
    345 KB · Views: 55
  • Unbenannt-5.jpg
    Unbenannt-5.jpg
    468.7 KB · Views: 57
Last edited:
Ok Guys just took my 11 Panel image and processed it further with "register match images" and then "Gradient merge mosaic".

Look what silly result i got... looks like modern art ;-P
And the second what it should be, left bottom would be no 12 which could never be added...
Edit added a third pic, this is the result with more but not exactly "standard values" during the "register match images" and then "Gradient merge mosaic" process, but also a very poor result on the areas where frames overlap -> not usable.

Seriously is there a working way to create a 12 panel mosaic in Pixinsight?
Please im no Pix Pro, I need a detailes tutorial (best video) which I could follow. :-(
I also created a large mosaic with a 50mm lens, from the same site!

Having had the same struggle with that 'last corner panel' that refuses to play nicely, I was motivated to create a new mosaic script. You will find it in SCRIPT -> Mosaic -> PhotometricMosaic. I have also written a detailed help file (page with a bent corner icon at the bottom left of the main dialog) that explains how to use it. This include a 'Quick Start Guide' and a 'Tutorial'. It also explains how to register the panels in a reliable way (Prerequisites section), and how to prepare them (Use the script 'Mosaic -> TrimMosaicTile').

I realize this means starting again, but having invested much effort in taking them, it is worth persevering. If, after following the advice in the help file, you don't get really excellent results, I will provide you with some help.

Regards
John Murphy
 
Provided that your images are plate-solvable, alignment should not be a problem. Background / gradient / colour matching is another matter!
 
Provided that your images are plate-solvable, alignment should not be a problem. Background / gradient / colour matching is another matter!
Provided that the images form a regular grid, and the data is linear and processes that don't conserve star flux have not been applied, PhotometricMosaic should be able to cope with strong gradients, and target panels will be color matched to the reference frame. Do give it a try!
 
I also created a large mosaic with a 50mm lens, from the same site!

Having had the same struggle with that 'last corner panel' that refuses to play nicely, I was motivated to create a new mosaic script. You will find it in SCRIPT -> Mosaic -> PhotometricMosaic. I have also written a detailed help file (page with a bent corner icon at the bottom left of the main dialog) that explains how to use it. This include a 'Quick Start Guide' and a 'Tutorial'. It also explains how to register the panels in a reliable way (Prerequisites section), and how to prepare them (Use the script 'Mosaic -> TrimMosaicTile').

I realize this means starting again, but having invested much effort in taking them, it is worth persevering. If, after following the advice in the help file, you don't get really excellent results, I will provide you with some help.

Regards
John Murphy


Hi John THIS sound great just overflew the manual and its overwhelming complex for me as a "standard user".

Question 1 where do I get the script? I have the latest Pix version but there is no such script in my scripts libary.
The only Mosaic scripts I have are "Mosaic Planer" and "Mosaic by coordinates"

Unfortunately after the stacking process Im already stuck just by reading your manual :-(
Use the script 'Image Analysis -> ImageSolver' to plate solve the mosaic tiles.
And the following steps -> I dont have any cluejust by reading :-(

Did you ever think about to make a youtube tutorial with your workflow?
This would be nice for Pix dummies like me.
 
Last edited:
Are you sure you don't have it. On my system (Windows 10 Pro; PI 1.8.8-7) it is in the scripts menu as indicated above (scripts > mosaic), or in the process explorer:
1620070598397.png

Use the script 'Image Analysis -> ImageSolver' to plate solve the mosaic tiles.
Do you have problems with this? Have you tried to run ImageSolver on at least one of your tile images?
 
Hi John THIS sound great just overflew the manual and its overwhelming complex for me as a "standard user".

Question 1 where do I get the script? I have the latest Pix version but there is no such script in my scripts libary.

Unfortunately after the stacking process Im already stuck just by reading your manual :-(
Use the script 'Image Analysis -> ImageSolver' to plate solve the mosaic tiles.
And the following steps -> I dont have any cluejust by reading :-(

Did you ever think about to make a youtube tutorial with your workflow?
This would be nice for Pix dummies like me.
The script should be located at:
SCRIPTS -> Mosaic -> PhotometricMosaic
or download the latest version (fixes a crash on MacOS):

The only parts of the help file you need to read to get started are the Prerequisites and Quick Start Guide.

The ImageSolver script also has help. It looks more complicated than it is.
  • Select the 'Active window'
  • Find an object that is approximately central in each image, and 'Search' for this in ImageSolver.
  • Enter your Focal distance (50mm) and your camera's pixel size
One day I might have enough time to create a youtube tutorial...

Regards, John
 
If you could upload two or three of your tiles to a cloud location and post a link we might be able to walk through the basic procedures with examples.
 
Thank you John and Fred I really appreciate your support :love:

Actually I managed to install the script and it works for now when I open it :)

I need some time to have a deeper look on it, but currently Im tired so it does not make sense today.

Just tried to work the script to get 2 panels together (without any Prerequisites ) and I receive an error that both images must have the same dimensions!?
I have croped every stack of my 12 piece a little with dynamic crop to get rid of the bad border areas, so does it no more work with my 12 stacked images?
Or is it because I didn't the Prerequisites and after that it will work?
 
Don't panic!
This simply means that you are trying to use two images that have not yet been registered to each other. This will not work.
You need to plate solve the two images, and then register them together using the Mosaic by coordinates script. Both images will then be much larger, exactly the same size, and have lots of black regions around them. Only after this step can you use PhotometricMosaic.
 
You will actually want to run Mosaic by coordinates on all of your panels at once. Each panel will then be the same size as the final mosaic.
 
One day I might have enough time to create a youtube tutorial...

Regards, John

I could send you my data and you could use them also to create a tutorial ;-)
it would for sure help thousands of astrophotographers.
Just an idea ??
 
Ok I tried to start from scratch step by step with the official Pixinsight reference.

I took the reference for imagesolver:

I filled the following values date, focal length, chip pixel size,
Everything else standard values.
After the script ran I received the following error message -> see pics.

So already a punch in the face on the first step :-(

Edit Tried also with Algorhytm "Polygons" -> Same errors
 

Attachments

  • Unbenannt-1.jpg
    Unbenannt-1.jpg
    255 KB · Views: 54
  • Unbenannt-2.jpg
    Unbenannt-2.jpg
    242.5 KB · Views: 69
Last edited:
Back
Top