How to create a multi-channel mosaic

I am seeing a lot of blochiness and uneven gradients in my mosaic - this does not appear to be easily correctable (or at least I have not found any way in APP to fix it)  Is there a suggested set of settings to resolve that?

Hi TIm,

Most likely, a lot of things that you refer to can be fixed/corrected in APP, but I will need to see the data. Can you upload an image or two showing the problems? And...

1) have you used a BPM in calibration?

2) in 6) integrate, have you used both MBB and LNC ?

3) Did you correct the individual mosaic panels for gradients first?

Kind regards,

Mabula

Here is a link to 4 files in an M42 mosaic green channel:

https://www.dropbox.com/s/3dpt9johqdu81o5/Archive.zip?dl=0

This is giving:

- I have already tried to use the 'Remove light pollution' on each, ( but they could probably be flatter)  For some reason the mosaic tool is adding additional gradients, and giving a strange distorted result.

My settings:

1) - no didnt do a bad pixel map.

2) in 6 -I'm using 1st degree LNC, and 3 interations, although I played with this and added more, but it didnt seem to help.

3) Yes I tried as best I could.

I have just retried this again, after quitting out and restarting.  The strange anomaly on the bottom right is gone, but the dark blotches remain.  I don't see these gradients in the separate files, but perhaps they are there but just not as visible when the files are viewed?

Posted by: Tim

Here is a link to 4 files in an M42 mosaic green channel:

https://www.dropbox.com/s/3dpt9johqdu81o5/Archive.zip?dl=0

This is giving:

- I have already tried to use the 'Remove light pollution' on each, ( but they could probably be flatter)  For some reason the mosaic tool is adding additional gradients, and giving a strange distorted result.

My settings:

1) - no didnt do a bad pixel map.

2) in 6 -I'm using 1st degree LNC, and 3 interations, although I played with this and added more, but it didnt seem to help.

3) Yes I tried as best I could.

 

Hi Tim,

I'll have a look at the 4 panels.

For some reason the mosaic tool is adding additional gradients

The actual mosaic registration will never alter the illumination of the data. A tool like LNC can however, because it tries to correct the illumination differences between the panels.

and giving a strange distorted result.

That will happen if dynamic distortion correction is unstable in registration. This happens usually when distortion correction is not really needed, but i'll try to figure out what's going on.

Will report back later today 😉

Mabula

@toc

Hi Tim,

Okay, first result with settings used and why:

Done with APP 1.057 which is about to be released.

3) Analyse stars

To be able to have good registration of your mosaic, we want to detect more stars than for a non-mosaic situation. So I increased:

• automatic minimum #stars target to 1000
• automatic maximum #stars target to 10000 ( the amount of stars in your frames isn't very big, this ensures almost all available stars are analysed and used)

4) Register

• change scale stop to at least 10 (this is needed for mosaics)
• use dynamic distortion correction
• disable same camera and optics
• registration mode to mosaic

5) Normalize

• set mode to advanced (for mosaics usually preferred)

6) Integrate

first I did an integration with default settings, only enable the normalization output map to see how data normalization is working.

I think that the initial distorted lower-right panel probably happened due to too little stars detected in 3) analyse stars with default settings. So for a mosaic, try to have more stars detected if registration is problematic. More stars means more registration points, making registration more robust. (On the other side, detecting more than 5000 stars in panels usually is overdone and will only slow down registration)

In a next post, I'll study the LNC + MBB correction on your data to make a nice seamless mosaic

Mabula

Hi Tim,

Yes, I see that correcting the panels with LNC + MBB is a bit problematic.

Chances are that this is due to your Light Pollution Correction (LPC) on the initial panels, so it is worthwile if you could also share the panels without LPC so I can have a look at those to verify if that works better.

Anyway, this is what I get with LNC 2nd degree 10 iterations and MBB at 15%:

We see your problem, the black area. We also see that the normalization map shows that normalization is problematic, we expect it to be smooth, but it isn't. Data normalization is not working. To me, this could indicate that the initial Light Pollution Correction is not done properly and has introduced new gradients that are very unnaturel. I can only confirm this if I can have a look at the panels before Light Pollution Correction. Can you share those?

Kind regards,

Mabula

Hi Tim, @TOC

Did you try integrating on the panels with LNC before correcting the panels with Light Pollution Correction?

Was wondering if that gave a more stable result 😉

Mabula