The Andromeda Galaxy - 3 Panel mosaic

[Stefan B]

Hi everybody,

after the Iris nebula I decided to go for a brighter target and picked the brightest I could find, the Andromeda Galaxy. Unfortunately, M 31 doesn't fit my FOV. At least, not nicely and it would look very cramped. So it had to be a 3 panel mosaic. Have never done a deep sky mosaic, but here is what I came up with:

It consists of three upright panels with a total integrated exposure time of 20+ hours. I had to tinker with it quite a bit. The first try was to just stitch the stacks right away in ASTAP. The autodev in StarTools revealed that there were different gradients in the picture and they were not very uniform in the three panels. This kind of surprised me since the data for the three panels was recorded in three consecutive nights with similar conditions, i.e. starting and ending the sessions happened around the same time. Moon shouldn't have been an issue.

Anyway, I couldn't get rid of the gradients within the mosaic so I went back to the stacks of the three single panels, cropped away the artefacts on the edges and wiped the data in order to get rid of the individual gradients. I saved the stacks without stretching them and keeped them linear. After stitching these wiped stacks the result was a monochrome mosaic without color information. After some thinking I noticed my mistake. Out of habit I opened the file via 'Open' and told StarTools that the dataset came from a OSC/DSLR. Obviously, using this option and saving the file before the color module results in black and white images.

Okay...opening the individual stacks again and chosing 'Linear' only. Cropping, wiping and stitching again. Long story short(er): I tried some different combinations of checking/unchecking of background equalisation and background merging within ASTAP. The best option for this data was no equalisation but background merging.

Unfortunately, my notebook is obviously underpowered so I had to bin 60%, since otherwise the machine repeatedly surrendered. Nonetheless, I am happy with the outcome. StarTools 1.8 is really a blast. The SuperStructure module is really an impressive piece. And I'd like to say that the Shrink module works great and really easy to use. Reducing the stars makes a huge difference on the final image. That's sometimes underestimated I think.

Feedback and CC welcome! Clear Skies!

Regards
Stefan

PS. See https://www.astrobin.com/d7950r/ for technical details.

[Carles]

beautifully done, Stefan!!

I've also wiped my images before stitching them, but not sure what stitching parameteres were in ASTAP,
will check next time.. but indeed very nice result for a 3 panel mosaic, can't really see where it stitches!

and the colours!!!

keep the good work!! and the good tips! hehe

IIIIIFFF I had something to pick on, would be the circle shape on the galaxy center. That's probably an "artefact"
of HDR-Tame. ? but that is looking at it very closely!!

Kind Regards

Carles.

[Stefan B]

Hi Carles,

IIIIIFFF I had something to pick on, would be the circle shape on the galaxy center. That's probably an "artefact"
of HDR-Tame. ? but that is looking at it very closely!!

Well spotted! That's one of the flaws I mentioned on Astrobin It does not come from HDR tame (I used Reveal All), but it came from Shrink. The automatic mask in Shrink mistook the galaxy core for a star (despite my praise for the Shrink module! ) and I've not seen that. So you can see the remnant of the mask there and the artefact it has produced. I only noticed after denoise etc. and I was too lazy to layer/blur it out. Thought it might be a nice easter egg for the pixel peepers - congrats for finding it!

Regards
Stefan

[Stefan B]

can't really see where it stitches!

Actually one shouldn't point out the flaws in own images but look for example below M 110. You can see that the stars and the background appear somewhat blurry. I guess that comes from stitching. The region corresponds to the edge of both panels.

[happy-kat]

Great image lots of colour captured.
I like to use HDR reveal DSO core rather than reveal all when it's a central bright area to coax from.
I've only done a couple mosaics but I processed each frame separately then stitched the final images.

[Mike in Rancho]

What a truly great image!

I will have to remember this post for your mosaic tips (I can't fit M31 in FOV either).

I can't see the stitching error you mention, but for that little ringing artifact you say is in the core (very faint, mind you), you ought to be able to mask it (or the whole core out to just past it), and then in Layer just add a tick or two to the filter kernel radius setting (and mask fuzz if even needed, but note the pixel scale is different) while zoomed in to watch the third panel. That might give a little blur to the artifact (if you don't want to go redo the point where it first appeared) so that it isn't noticeable. Also a good trick to tame an overly bright central core.

[Stefan B]

Great image lots of colour captured.
I like to use HDR reveal DSO core rather than reveal all when it's a central bright area to coax from.
I've only done a couple mosaics but I processed each frame separately then stitched the final images.

Thanks!

Have you shared the mosaics anywhere? I'd love to see them. I also considered processing the panels separately but worried that I would not be able to process them in way similar enough to not look weird after stitching. So I thought stitching linear data and processing the mosaic was a safe way to get a decent and homogenous result. How did you for example ensure that you stretch the stacks in a similar fashion?

What a truly great image!
... but for that little ringing artifact you say is in the core (very faint, mind you), you ought to be able to mask it (or the whole core out to just past it), and then in Layer just add a tick or two to the filter kernel radius setting ...

Thanks for the feedback and the tip, Mike! I might give that a try

Regards
Stefan

[Stefan B]

I like to use HDR reveal DSO core rather than reveal all when it's a central bright area to coax from.

Sorry, missed that in my previous post. I usually also use rather Reveal core but Reveal All did a pretty decent job here in exposing the details all over the galaxy. So I went with that.

[happy-kat]

It was a multi pane image of orion area using camera lens 85mm, it's on SGL.
Using decon I sometimes coax more from the middle with a selective mask, and do several steps in sharp module using different masks but this is only on bright core areas as I'm wary of art over reality.
I used the logfile to manage similar processing and ICE to stitch

[Stefan B]

It was a multi pane image of orion area using camera lens 85mm, it's on SGL.

I wasn't able to find it. Do you have a link?

Using decon I sometimes coax more from the middle with a selective mask, and do several steps in sharp module using different masks but this is only on bright core areas as I'm wary of art over reality.

I usually try to avoid selective masking. It's too much fiddling for me And I am usually happy with the presets and some tweaks working on the whole image (without masked out areas using the automasking features).

I used the logfile to manage similar processing and ICE to stitch

I can imagine that this works fine for HDR or Sharp etc. But how do you stretch your different panels? With AutoDev? Let's take M 31. One panel will have the extremely bright core and another panel will have the fainter peripheral structures. You will have different ROIs since the panels contain different structures. Wouldn't you come up with incompatible stretches? Or do you use FilmDev and stretch with identical parameters? I am really interested to know

Regards
Stefan