Issue with Deep Sky Stacker and wide angle shots

[kalon]

***WARNING*** Linked photos can be HUGE. Check your Wifi!

I'm having a challenge stacking wide-angle shots (Fujifilm X-Pro1 with a Samyang 12mm f/2.0 lens) using Deep Sky Stacker. When I use large numbers of exposures, the final image is blurred, even minorly out from the centre, and the extremeties of the shots even have doubled stars. I've tried fumbling around myself, to no avail, so I appeal to you, the experts...

I have two nice series of 45x15" and 45x10" shots taken at Hargraves Lookout. Each individual image is relatively crisp all the way to the edges (just a single capture - see https://www.flickr.com/photos/92167713@N00/30947334507/sizes/o/ )

If I stack just 4 of them, I get a decent result (stack of 4, processed - see https://www.flickr.com/photos/92167713@N00/30894667847/sizes/o )

However, when I stack all 40+ of them in DSS, I get a muddy, smudged, doubled result in the corners, and I can't seem to change it... (cropped processed stack of 40+ - see the bottom portions of https://www.flickr.com/photos/92167713@N00/30947396737/sizes/o



Uncropped, unedited stack, see https://www.flickr.com/photos/92167713@N00/45837007052/sizes/o )

To me, it seems that the fish-eye effect of stars at the outer edges that is reduced as the stars move in to the centre/out of centre of the frame is not being corrected by software, leading it to the apparent motion of the stars smudging them, and eventually DSS realises/calculates that it is a "new" star, hence the doubling. While not all were immediately taken one after the other, all shots were taken with only a short delay between each capture - 5-10 seconds gap would be accurate, and only a handful longer than that between captures.

Can anyone make suggestions for the settings I should use? Or if it is a limitation of the software, what alternatives there are to it?

[bojan]

Perhaps ypou should manually select stars (from image centre) for registration ("comet" mode).


It seems the problem is lens coma at corners of the frame, which affects registration.



Could you try with smaller aperture (higher f/ number)?

[kalon]

Thanks for the suggestions, will certainly try those!


I don't really want to waste these shots, but I will try a batch with a higher f stop on my next outing.


Any other suggestions?

[skysurfer]

Or if you have Photoshop, you can load all frames in Adobe Raw and 'fishing' them by 'Filter => Wide angle ...' to make fisheye like images of the rectilinear images which indeed distort strongly in the corners with such a wide angle.

[bojan]

I don't think the problem is fish-eye distortion here - if there is one, all frames will be distorted in the same way, and if tracking was correct, stacking should not be the problem like we see here.

But, due to coma the star images in the corners are not round, DSS have trouble calculating the centres for tose stars- and the registering and resulting stack can't be accurate.

[kalon]

Thanks for the replies! I have found the issue.


Going back to fundamentals, I re-read some tutorial pages from different people on the web and followed some of the recommended settings, which resulted in me changing the star detection threshold from 80% to 10%.


The number of registered stars jumped from ~450 stars to ~5000 stars. This identified more of the outer stars that move in/out of frame, and allowed them to more consistently be moved as needed, and has corrected this issue.


Edit: The stack now takes significantly longer, but that's the trade off, and I can live with it given I can now use all of my images!

[bojan]

Excellent!
Glad you solved it ! :thumbsup:

[kalon]

The finished product (post-processing-heavy):
https://www.flickr.com/photos/92167713@N00/44084433630/in/dateposted-public/

[bojan]

:thumbsup:Very nice