Alignment issues with MosaicByCoordinates

[deepanshu29]

I am having issues with building a mosaic. Overall mosaic plan is total 15 panels but as of now I only have data for 2 panels. I was trying to build mosaic with these two panels for practice. My flow was -

Calibrate > Debayer > Subframe Selector > Star Alignment -> Integrate -> DBE -> Plate Solve -> PCC -> MosaicByCoordinates -> GradientMergeMosaic

I tried multiple different Pixel Interpolation methods but I got the same issue pretty much with all of them. Stars on edge of the frame do no align. Please see the image below.



Entire frame -



Initially I registered individual images for each panel with Thin Plate Splines and Distortion correction checked but with that I saw even bigger errors down the line. So I went back and registered again using Projective Transformation.

I believe I am doing something wrong but unable to figure out what. Any help is appreciated.

[deepanshu29]

FYI, images were collected using a full frame DSLR (Canon 6D mark II) on Skywatcher Esprit 100 with 0.75x reducer (415mm focal length, f/4.15), if that matters.

Also, I left all boxes unchecked in Mosaic Geometry, and let it fill automatically by 'calculate' button.

[John_Gill]

Hi,

Have a look at this thread on the forum from a few days ago. "Mosaic of a mosaic not aligning correctly..."  maybe you have a similar issue.  I followed Juan's instructions and the alignment worked perfectly.

space is not black
John

[Andres.Pozo]

Did you activate the distortion correction? If not, try again with it activated in ImageSolver.

[deepanshu29]

Distortion correction was on in ImageSolver but I left it at default settings. Do I need to use specific settings?

It does seem like that there is non uniform distortion on extreme edge of the frame. I am a bit clueless on how to fix that.

[Andres.Pozo]

Distortion correction was on in ImageSolver but I left it at default settings. Do I need to use specific settings?

It does seem like that there is non uniform distortion on extreme edge of the frame. I am a bit clueless on how to fix that.

MosaicByCoordinates depends on a good astrometry for all the areas of the image. It is important to activate the option "Generate residuals image" and check that the zones that overlap other tiles in the mosaic are well calculated. If there is any displacement between the predicted position of the stars and their real position then the tiles will not match.

The latest version of ImageSolver has a new option "Advanced alignment" that tries to improve the alignment in the corners of the image. It is slower but perhaps in your case could make the difference.

[deepanshu29]

Thanks Andres,

I will give these options a try and also generate the residual image to see if stars in corners are calculated correctly. I will report back on how it goes. Thanks again.

[deepanshu29]

I am still having serious issues with assembling mosaics. I tried ticking "Advanced Alignment" option in Image Solver that seemed to have helped but it did not completely fix the alignment. I also generated distortion map but I don't know what to make out of it.

It seems like my optical train does have some weird distortion (It is SW Esprit 100ED with TS 0.75x reducer, Canon 6D Mark II).

I have uploaded the stacks of two panel that I am trying to assembling to google drive and also the two individual subs (calibrated and debayered) from each panel. In case if anyone wants to take and look and help getting the tiles align correctly.

Here are the files -  https://drive.google.com/open?id=1acdvPZ7jppvHMbHarLDbxMV3TZ1aoHjt

Unless I figure out how to assemble the tiles, my efforts on collecting data are rendered useless and I might have to look for a different optical train that does not have this serious distortion. Any help is seriously appreciated. Thank you in advanced.

-D

[Andres.Pozo]

Hi,
your images have distortions at the corners, but I think that they are quite normal, I have seen much worse.

In order to get a good astrometry for these images, you have to increase the magnitude of the catalog. The images are deep and the automatic values for the catalog and magnitude are too conservative. You can use the PPMXL catalog and an magnitude about 12.5.

You should also update the version of PI. The very latest version published yesterday (1.8.8) has changes that should improve the behavior of ImageSolver.

This message has attached a test with your images. I cropped the result to show the overlapping part of the mosaic. I can not seen any geometric defects. The only defects are the artifacts introduced by GradientMergeMosaic around the stars in the borders of the tiles.

Andrés.

[deepanshu29]

Hi Andres,

This is great, thank you so much for having a go at my images. Yes, I did not change the star magnitude. I remember adjusting the star sensitivity to -1.2 but did not touch anything else.  I will go ahead and try your suggestions and also update the PI to 1.8.8

Thank you once again.
-D