Simple mosaic results in swirled areas
I tried dabbling with creating a Mosaic following a video tutorial provided on your site. Excellent tutorial but results were very far off.
So I took a different route, processed each of the two frames separately into their own respective rgb images without any stretching, just created and saved.
Then I used these two files as the lights for a Mosaic, setting all the recommended settings in the video. They actually got assembled opposite of target center! So I started trying different settings. Changing one thing at a time till at least I got the center of the object in the center! My files were of the M31 Andromeda galaxy, shot during a gibbous moon to test the mosaic function.
While I did get the center correct, the outer edges of the galactic star disk were rendered as swirls! And this was not only significant but would not stop happening regardless of any other changes I made.
Rather comical actually.... and probably completely noobish in some error.
I did do some searching here in the forums for swirl, didn't find much, so asking now. What could be causing this. I am attaching a simple screenshot of the affect to keep file size down.
Did you ever figure this out? Cuz I got some swirls, too, and I don't quite understand it
Not sure how many mosaic panels you are trying to align but I guess you are working with more that a 2 x 2 mosaic.
To me it seems APP has not been able to achieve a strong result while pattern matching certain of your panels to their neighbours and made a wrong choice as to the correct alignment.
Dynamic distortion correction has then stepped in and grossly distorted some of the panels so as to achieve a 'good fit' based on this wrong choice.
I would assume that you have already tried the usual recommendations for achieving a good fit in Star registration, specifically increasing star detections, extending the scale start/stop range and changing the pattern recognition choice to eg triangles?
I wonder what would happen if you were to try first aligning just local pairings of your panels from the worst distorted parts of yourmosaic, preferably I think without using dynamic distortion correction if possible.
If you can get a good alignment between adjacent pairings of individual panels I think you should be able to gradually build up the full mosaic by successively integrating larger units of the picture. Basically if you can achieve larger units of correctly matched sections, by the time you come to the final integration, pattern patching will hopefully have sufficiently large overlapped areas between the separate sections so as to make the correct decision with regard to the overall alignment.
I appreciate that this approach creates more work on your part but hopefully will eventually give an acceptable result.
For your future mosaics you perhaps need to decide what aspects of your current data might be causing pattern-matching to fail, for example insufficient overlap between panels, very noisy data and/or badly distorted stars in frame edges.
Clear skies and happy processing
@rawhead Do you first process the individual panels and then try to create the mosaic or do you load all raw files and let APP chew at them? What settings do you use in tabs 3 to 6 for creating the mosaic?
Thanks for the input; last night I did try changing some parameters and it appears changing the "scale start" and "scale stop" seemed to do the trick. I had them set lower because my overlaps were greater than 10% (15% to be precise). I am now working to get rid of the cross-shaped dark band in the middle (this is s 2x2 mosaic).