Help with mosaics that have small overlap

[pluckas]

I'm experiencing mixed success trying to create mosaics from image pairs that have only a small overlap using Star Alignment. The error is always: "Unable to find an initial set of putative star pair matches".

I've tried lowering the Log (Sensitivity) and raising Peak Response to their limits, as well as manipulating some of the other parameters without success. I've also tried defining previews around the stars in common. In that case the routine can find up to 4 or 5 stars - but never the minimum of 6. Frustrating to say the least - especially as there are at least 12 visible to me.

The images are rotated with their 'overlap' correctly aligned, and have at least a dozen starts visible in both. They can be combined quite successfully by hand in other applications.

Any advice on how to deal with small overlap image pairs would be gratefully received.

Paul

[georg.viehoever]

If everything fails, try DynamicAlignment (the manual brother of StarAlignment).

[pluckas]

Thanks Georg,

It seems to work, but I'm not quite sure of the correct process after producing an aligned version of one of two image pairs. A quick search for help on using the dynamic alignment tool for mosaic creation has yielded little.

See attached screen shot. What are the next steps for creating a mosaic of the two frames now that one of them is aligned?

[georg.viehoever]

Hmm, first of all it seems that you do have considerable overlap, so I wonder why StarAlignment failed. Did you define one preview in each of the two images to indicate the overlap regions?

- If you can get StarAlignment working, try this: http://pixinsight.com/forum/index.php?topic=1563.0
- If you stilll have seams, try GradientMergeMosiac http://www.youtube.com/watch?v=H2r3ZYqCaP8&gl=US&lr=1,http://pixinsight.com/forum/index.php?topic=4105.0, http://pixinsight.com/forum/index.php?topic=4105.0
- If you need to use DynamicAlignment:
-- Enlarge your reference frame to the final size of the mosaic by adding borders (Crop tool in Geometry category)
-- Align your second frame to it
-- Combine both frames using ImageIntegration or GradientMergeMosaic

Note: Judging from your screenshot your source frames have a number of defects. Of course these still are there after creating a mosaic...

Georg

[pluckas]

Thanks again Georg.

The only way I have managed to merge the R, G & B components of this mosaic is by using DynamicAlignment. Curious, as the L merger worked quite well with StarAlignment once I had tweaked the sensitivity and peak response values thanks to the advice from Juan. Other than filtered, the only difference between the two is that the RGB were acquired at Bin2 - otherwise the overlap was much the same (about 5%).

The column defects (amplified by the fact that the individual exposures were dithered) were not effectively treated during integration, as I did not spend time on on pixel rejection (this has been an exercise on using the mosaic features of PI).

I'm reminded of why I have not spent much time on LRGB mosaics over the years. It is a very time consuming process - even with a well organised workflow.

Thanks for your excellent GradientMergeMosaic tool.

Paul

[Juan Conejero]

Hi Paul,

The only way I have managed to merge the R, G & B components of this mosaic is by using DynamicAlignment.

This should not happen. StarAlignment should be able to solve this problem. Could you upload the images somewhere?

[pluckas]

Thanks Juan. I've placed them in a dropbox folder - I'll email you the details.

Paul