Cygnus Loop - 4 panel mosaic

User images created with StarTools.
bobharmony
Posts: 20
Joined: Fri Jan 22, 2021 2:42 am

Cygnus Loop - 4 panel mosaic

Post by bobharmony »

I am attempting my first mosaic and chose the Cygnus Loop as my target. I had great plans for getting hours of data for each panel, but the weather intervened and I didn't get my first session until September, instead of July as I had hoped to. As a result I have 50 minutes of subs for each panel.

I stacked each panel using DSS and processed them individually in StarTools, then stitched them together using MS ICE. I tried to be very careful about making sure there were no variations between the panels, but wasn't entirely successful in doing that. I like using AutoDev and selecting a ROI when I process, but that may be a bit problematic for this type of processing. I went ahead with it and got this result:
Veil-50-minutes_stitch-shrink.jpg
Veil-50-minutes_stitch-shrink.jpg (473.31 KiB) Viewed 4495 times
After more research I came across Stefan B's description of his 3 panel M31, in which he used ASTAP to stitch the panels in a linear state before taking the panels into StarTools and am in the process of trying to follow that process now. My first effort took my 4 280MB panels and created a 3.3GB file with lots of black space around the actual mosaic data. I could barely get it into StarTools to look at it and just trying to crop away the excess space around the outside of the image resulted in an out of memory condition.

Having read Stefan's post more closely, I see I need to modify each panel stack as he describes and then go to ASTAP for stitching, then to StarTools for processing. I will post further in this topic as I progress. ATM I am running another stitching attempt where I specified merging the backgrounds in ASTAP, I'll see how that goes as well.

Bob
Scope: C6N
Cameras: Canon 60D (unmodded), Nikon D5300 (modded)
Stefan B
Posts: 422
Joined: Sat Oct 31, 2020 8:59 pm

Re: Cygnus Loop - 4 panel mosaic

Post by Stefan B »

bobharmony wrote: Sat Oct 23, 2021 3:42 pm My first effort took my 4 280MB panels and created a 3.3GB file with lots of black space around the actual mosaic data. I could barely get it into StarTools to look at it and just trying to crop away the excess space around the outside of the image resulted in an out of memory condition.
I also had this problem! But after astap is done with stitching it shows you the result with the black space around it. You can already crop the image here and greatly reduce the size before going into Startools. I think you can crop by drawing the cursor while holding the right mouse button.

Thanks for sharing your results!

Regards
Stefan
bobharmony
Posts: 20
Joined: Fri Jan 22, 2021 2:42 am

Re: Cygnus Loop - 4 panel mosaic

Post by bobharmony »

Thanks for the tip on cropping (and the "how to"), Stefan. I was able to get the image down to about 1G, which StarTools and my PC can handle.

So far my experiments with ASTAP have not gone well. The overlap areas have some slight offsets that produce double stars, makes me think I am looking at a 3D image without the glasses. Could be my issue as I don't have quite correct spacing for the CC yet with the new camera, so my edge stars don't looks so good.

This did not happen with MS-ICE when I stitched the processed images together, but ST had done a fine job of cleaning up the stars in the Deconvolution module.. I may see if there is a way to play with rotation of the panels in ASTAP as there may be an issue there. I will also take my 4 processed files into ASTAP and see if they are stitched cleanly, that may give me an indication if rotation is an issue.

Work continues, nothing useful to report yet.

Bob
Scope: C6N
Cameras: Canon 60D (unmodded), Nikon D5300 (modded)
bobharmony
Posts: 20
Joined: Fri Jan 22, 2021 2:42 am

Re: Cygnus Loop - 4 panel mosaic

Post by bobharmony »

I think I figured out that ASTAP does have a rotation issue (or my data is very tough to work with) when trying to stitch multiple panels that aren't in a 1xn configuration. I am doing a 2x2 mosaic with the panels laid out as:

1 3
2 4

To work around the issue I did three stitches; first ! stitched panels 1 and 2, then I stitched 3 and 4. Finally I stitched the two results together to make up the completed mosaic. I have opened the result in StarTools and the resulting image is much cleaner at the edges than my attempts to do all 4 panels in one pass had been.

Now I am ready to try Stefan's method for running each of the panels through wipe, attempted to get them as even matched as possible, after which I will stitch them together in ASTAP and take the whole enchilada into STarTools for processing. This is what progress looks like to me for today!

Bob
Scope: C6N
Cameras: Canon 60D (unmodded), Nikon D5300 (modded)
Mike in Rancho
Posts: 1141
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: Cygnus Loop - 4 panel mosaic

Post by Mike in Rancho »

Looking forward to the results! :thumbsup:

I have yet to try a mosaic of my own. For processing, I have only tried the two-panel that Eyal posted in the big Elf practice thread. Of course, it was a mono cam and each of the panels had Ha, R, G, and B. Ugh.

I gave ASTAP a shot, but unfortunately each of the 8 panels need to be normalized/gradient cleaned to match, before any stitching, or else the seams were blatant.

In the end I "cheated," processing the two full panels in ST independently but as closely matched as I could eyeball them, then manually stitched them together in Gimp using a gradient mask blend. :confusion-shrug:
bobharmony
Posts: 20
Joined: Fri Jan 22, 2021 2:42 am

Re: Cygnus Loop - 4 panel mosaic

Post by bobharmony »

Mike in Rancho wrote: Tue Oct 26, 2021 6:33 am Looking forward to the results! :thumbsup:

I have yet to try a mosaic of my own. For processing, I have only tried the two-panel that Eyal posted in the big Elf practice thread. Of course, it was a mono cam and each of the panels had Ha, R, G, and B. Ugh.

I gave ASTAP a shot, but unfortunately each of the 8 panels need to be normalized/gradient cleaned to match, before any stitching, or else the seams were blatant.

In the end I "cheated," processing the two full panels in ST independently but as closely matched as I could eyeball them, then manually stitched them together in Gimp using a gradient mask blend. :confusion-shrug:
Hi Mike

Check out Stefan B's post on his M31 mosaic. He proposes a way to clean up the gradients in Wipe before stitching while maintaining the data in a linear state, so the entire view can be stitched before loading into StarTools for "normal" processing. That is what I am trying to recreate here. So far I have just been testing to make sure ASTAP can stitch the stacked tiles together.

My attempt in the first post in the thread did what you described, processing each stack individually in StarTools trying to get them as closely matched as possible before stitching them together. The result isn't bad, but it is not the seamless image I would like to wind up with. There are some weird background color variations that I would like to clean up, among other things. I will also be experimenting to see if ASTAPs background equalization could be a useful part of the process, although I am getting a little ahead of myself with that thought.

Bob
Scope: C6N
Cameras: Canon 60D (unmodded), Nikon D5300 (modded)
Stefan B
Posts: 422
Joined: Sat Oct 31, 2020 8:59 pm

Re: Cygnus Loop - 4 panel mosaic

Post by Stefan B »

Just a short notice: I had to bump up the dark anomaly filter until further increasing had no more impact on color. Luminance didn't change above DAF of 3 so I went with that initially. But the stitched panels had a somewhat different coloring while no more luminance gradients were visible. Bumping up DAF to 12 or something like that helped without having negative effects. Apart from taking longer to calculate.

Regards
Stefan
bobharmony
Posts: 20
Joined: Fri Jan 22, 2021 2:42 am

Re: Cygnus Loop - 4 panel mosaic

Post by bobharmony »

After some trouble trying to get ASTAP to platesolve my tiff files out of StarTools I had one of those aha! moments and realized I should try to use MS-ICE to stitch the tiff files, which worked like a champ. To review, I am attempting to even out the background variations that were present in my first attempts at putting the mosaic together. I am using a method Stefan B documented in more detail in his Andromeda post in the Gallery. The summary of steps used in the process I adapted from Stefan's work:

Stack the individual panels using DSS with Ivo's recommended DSS settings.
Load each panel into StarTools selecting the "Linear" option, bin 50%, crop artifacts, and run the Wipe module.
Save the resulting file as a .tif.
Load the 4 panels into MS-ICE and stitch them together.
Load the stitched result into StarTools using the "OSC" option and process normally. In this case I ran Autodev, Wipe, AutoDev with ROI, HDR, Deconvolution, Color, Shrink, and final noise reduction.

I think the result is much better balanced than my original effort which you can see in the first post of this topic. This was a very rough cut to see if the process would work and produce usable results. I will go back and fine tune the whole thing now that I have validated the process Stefan developed.
Veil-P1-4-50-minutes-wiped_stitch-ST-web.jpg
Veil-P1-4-50-minutes-wiped_stitch-ST-web.jpg (453.77 KiB) Viewed 4342 times
Bob
Scope: C6N
Cameras: Canon 60D (unmodded), Nikon D5300 (modded)
Mike in Rancho
Posts: 1141
Joined: Sun Jun 20, 2021 10:05 pm
Location: Alta Loma, CA

Re: Cygnus Loop - 4 panel mosaic

Post by Mike in Rancho »

Cool! Hard to tell though, it's reversed and I think a bit less stretched?

I went hunting for my folders where I did that mosaic, turns out February so a good ways back. I know I have a yellow pad with notes all over it too, but that'll be hard to find right now (I have lots of yellow pads lol).

But I was able to reconstruct a few things by looking at the interim files in various subfolders. The original FITS (from Eyal/Imtl on CN) were stacked in PI. I actually used Gimp, not ST, for my strategic cropping of these FITS files prior to doing mosaic in ASTAP. Though I may very well have AutoDev'd them in ST first in order to help get the precise pixel/resolution numbers for those crops, then took those numbers over to Gimp, which can save in FITS.

I gave it a re-test just now and ASTAP still stitched them together just fine. Albeit it was a huge nearly 1GB file surrounded by massive black space that would itself then need major cropping. And while ST (on my laptop) did take a good long while thinking about opening this 1GB FITS, it eventually did so and even managed an AutoDev, which looked pretty normal to me.

Gimp can scale also, if necessary. I think I note you say you used bin in ST for some of the pre-stitching prep.

Have not yet tried out an ST-saved tiff to see how that might work/not work as two of you are now saying. ASTAP is famous for turning everything into FITS files, and I suppose it does that here too? That said DSS is also handy for doing TIFF to FITS conversions, and allowing various types (16/32 bit), in case that is what is throwing off the results. Unsure, would have to test.
bobharmony
Posts: 20
Joined: Fri Jan 22, 2021 2:42 am

Re: Cygnus Loop - 4 panel mosaic

Post by bobharmony »

After thinking about the process I realized that I had applied the Wipe module to the image twice, once when preparing each panel for stitching and again to the completed mosaic. My recollection is that this is not a recommended flow.

With that in mind I took another stab at process the stitched result, this time starting at the "AutoDev with ROI" step and moved forward from there. I like this result the best of what I have done so far. There are a couple of odd background spots showing up, but I am going to take a break for a while and leave it alone.
Veil-P1-4-50-minutes-wiped_stitch-rotated-ST-one-Wipe-small.jpg
Veil-P1-4-50-minutes-wiped_stitch-rotated-ST-one-Wipe-small.jpg (466.74 KiB) Viewed 4134 times
Comments and questions welcome.

Bob
Scope: C6N
Cameras: Canon 60D (unmodded), Nikon D5300 (modded)
Post Reply