Continuing problems with Mosaics.

[Robert Q. Kimball]

Hello,

I am fairly experienced with PixInsight but continue to have problems with building a mosaic from four panels.  Earlier a participant, pfile, gave me some great advice, but I still can't get my a mosaic from my data.  I have watched "Creating a Seamless Supermosaic, by Steve Allan" on YouTube many time and completely understand it.  I just can't get it to work. 

First, all of my four panel can be registered in pairs, ( ie. 1+2, 2+4,4+3, and 3+1)  there is no problem with star registration.

Here is what happen. Each time I add a panel to the mosaic, the added panel is "redder" than the first panel.  by the time I add the fourth panel the image is very, very red.  Each added panel seems to have a more blurry noise background.

Here are links to the four panels.  Any help would be appreciated. 

https://www.dropbox.com/s/2g33nlxlo8axb69/Heart_Panel_1_DBE.xisf?dl=0
https://www.dropbox.com/s/14gy55gg4t895p0/Heart_Panel_2_DBE.xisf?dl=0
https://www.dropbox.com/s/bobrfc8tlfvyoty/Heart_Panel_3_DBE.xisf?dl=0
https://www.dropbox.com/s/8x710o300nnx7ro/Heart_Panel_4_DBE.xisf?dl=0

Bob

[pfile]

i successfully made a mosaic from the data. i first used the dnaLinearFit script to match the brightness of each frame. 1 as reference to 3 and 2, then 3 to 4.

i then used staralignment in mosaic mode to build the mosaic; 3 as reference to 1, then the combined as reference to 4, then the combined as reference to 2. it turned out that the dnaLinearFit was not perfect, so i turned on 'frame adaptation' in star alignment and did it all again.

p.s. there is some problem with panel 4 - horizontal streaks in the stars. there must have been a bad frame in the integration.

rob

[the xisf is uploading, i'll edit the post when it's done.]

https://drive.google.com/file/d/0B6cW4ae2KyMJNXVrNUdzX0NEdEU/view?usp=sharing

[Robert Q. Kimball]

pfile,

Thank you again.  I'll see if I can duplicate your effort.  I'll figure out dnaLinearFit!  Now I owe you two beers in Las Cruces!

Bob

[Robert Q. Kimball]

Feeling like an idiot.  How do I get dnaLinearFit?

Bob

[pfile]

its a script written by david ault:

http://trappedphotons.com/blog/?p=994

middle of the page, under "intensity matching"

dnaLinearFit might not be necessary - did you use frame adaptation when you made your attempt? i didn't try skipping the dnaLinearFit step so i'm not sure if it helped SA get a better result when doing frame adaptation.

rob

[gvanhau]

Hello Bob
A different aproach would be:
Generate a star field the size of your whole mosaic, then register all the tiles to if.
finally use gradient merge mosaic to combine the registerd tiles.

Regards
Geert

[Robert Q. Kimball]

Hello Experts!

I still can't get a good mosaic from my panels.

I have put together matrix images before without any problems.  I think my problem has to do with my use of SGPro as my capture program.    When I use Artemis my debayer matrix was BGGR.  I was surprised that when I switched to SGPro the matrix changed to GRBG.  Whether this has any relationship to my mosaic issue I don't know.

I have tried to simplify my issue with new, better data.  I appreciate the help I've already received for Rob (pfile) and others.

Here are two panels each the integration of 4 X 10 min images with an Atik 3833L OSC camera.  I have carefully used DBE, Background Neutralization and Color Calibration to get them very similar in intensity.

https://www.dropbox.com/s/9d4im49ruamhmgm/Panel_1_DBE.xisf?dl=0
https://www.dropbox.com/s/rg6n6irxecj7tpv/Panel_2_DBE.xisf?dl=0

Then I did a Star Alignment using Mosaic/Union.  If you look closely at the resulting mosaic, the two images are very different in "background texture".  This didn't happen when I did the same type of combination with images captured using Atik's capture program Artemis.

https://www.dropbox.com/s/ol2lcmf4030lfmg/Mosaic.xisf?dl=0

The more panels I add the "creamier" the background gets.

I have also tried the Gradient Merge tool on this data and got terrible results.

Any help is welcome.

Bob

[chris.bailey]

Strange, I get very different results from using using the parameters below, the result is pretty much seamless??

I find Star Alignment - Union/Mosaic with Frame Adaptation and Distortion Correction works pretty well for two panel mosaics. Its when you get to 4 and above that more complex methods are called for.

Chris

[Robert Q. Kimball]

Thanks for your reply.  It does look a lot better with 2-D Spline.  I hadn't tried that.

Bob