RCC - M51, 4 hours from Bortle 5, is there more possible? Requests for constructive critique · Mina m.b. · ... · 22 · 909 · 15

Daniel Arenas:
Hi Mina,

So in that picture, you haven't used any kind of filter at all? I say that to try to reach some kind of explanations about halos.

Clear skies!

andrea tasselli:
Mina,

You only have a 80mm ED doublet so no surprise with either the issues with bright stars' haloes and the general poor dynamic of the image (and no Ha, this is hard to come by without Ha filter and even with that it isn't that simple). Put simply: you cannot afford the image scale you're imaging at. And 4 hours of RGB data is simply not enough, at all. Think 3 times that and you'll be in the right ballpark.

On the processing side you are overdoing it in more ways than one. Rather than go and deconstruct the process I'd rather have a go at it, if you're willing to share the stacked image and I shall share the processing with you.

andrea tasselli:
Yes, you got that right. When imaging at full broadband, as you are wont to do, than the issues with doublets can come back and bite you. If you were imaging with NB filters than the issues would be lessened and so if the image scale was more in keeping with capability of 80mm scope. To be honest I'd have already bought me a TS 6" f/4 Newt and be done with it since it is small and light enough to go with your existing mount.

Phil Creed:
I dunno.  Did you have any thin cirrus passing through during imaging?  That might cause some haloes.

I think something might have gone awry in the processing.  I shoot from a Bortle 6 so if anything, my skies are worse.  But for reference, I got this after 3 hours on a somewhat moonlit night:

https://www.astrobin.com/a3tu2v/

Similar speed (f/5.4 vs. your f/5.6) and somewhat similar focal length (540mm vs. your 450mm)

So something tells me it's the processing, not the data...which isn't the worst thing.  Precisely the opposite, because if the underlying data is good, it's very fixable.

The data looks a bit "flat", with some loss of contrast particularly around the spiral arms.  On the plus side, the cores aren't blown out and the dust detail near the center of both NGC 5195 and M51 look nice.

If it helps, here's my general workflow for a OSC camera:

Stack in Deep Sky Stacker (yes, DSS.  I'm THAT "old-school"...)
Dynamic Crop
DBE
BlurXTerminator, default settings
Photometric Color Calibration (can't get SPCC to work for some reason)
NoiseXTerminator, 0.70 overall / 0.15 on details
Histogram transformation (can't get GHS to work, either) until the histogram peak is ~0.20
NoiseXTerminator, ~0.50 / 0.15 (less aggressive round, if needed at all)

Then the "fun begins":

Rename image to "glow"  (or whatever suits you)
Starnet++ V2, rename stars-only image "stars" (or something else)
Adjust color saturation on "stars" image, set it aside.

"Glow" image:
Extract Luminance channel
Using Histogram transformation, slightly stretch it and then clip (yes, clip) the left side until ~45% to 50% of all pixels are clipped.  Use preview if necessary.
Select the newly-created luminance mask but *invert* it and find the general background illumination (mine's often ~0.20 to 0.24).
Using Curvestransformation, use the RGB/K option and pull the background point down to where you want it.  I generally prefer ~0.08.

Now *remove* the mask inversion.  Now you're working on the starless image but only the galaxy.
Adjust galaxy saturation to your liking under Curves Transformation.
Using the HDR Multiscale Transform tool to prevent core blowout and enhance the dust.  On this, I generally put the checkmarks on "to lightness" and "lightness mask" and select 5 or 6 layers and then play around with the "overshoot" function until it looks right.  Very trial-and-error.
I then do *some* unsharp masking.  Typically use something less aggressive than the default settings.

Using Pixel Math with the above image names, the expression is simply:
glow+stars
I usually name it something like, "combined" for the output image, but to each their own.

Hope this helps.

Clear Skies,
Phil

andrea tasselli:
Right off the bat you have a very sweet flat background, so nothing to complain about it. And very little data depth, as expected. Might I suggest you start using PI for pre-processing too?

• starXterminator
• DBE (remove large gradients as per Adam Block here: https://www.youtube.com/watch?v=7ldbGxFi-OM)
• DBE again in starless image
• Add images together again with PixelMath
• SPCC
• EZ Soft Stretch
• BlurXTerminator

var P = new BlurXTerminator;
P.ai_file = "/Applications/PixInsight/library/BlurXTerminator.2.mlpackage";
P.correct_only = false;
P.correct_first = false;
P.nonstellar_then_stellar = false;
P.lum_only = false;
P.sharpen_stars = 0.30;
P.adjust_halos = -0.25;
P.nonstellar_psf_diameter = 0.00;
P.auto_nonstellar_psf = true;
P.sharpen_nonstellar = 0.90;

• Extract L
• LRGBCombination with L to saturate RGB image (2 times with saturation set to 0.310 and 1 time with saturation to 0.310 AND Chrominance noise reduction)
• RangeSelection to create mask of galaxy
• add to L layer and protect all BUT galaxy
• use HDRMultiscaletransform  and then LocalHistogramEqualization to sharpen galaxy core - test different variations etc.
• apply L layer back on RGB with RangeSelection and all settings default.
• From extracted L layer create starmask again (once the mask is created in a new file - add the stars back to the L layer again (back arrow on the L view))
• Make stars slightly bigger (MorphologicalTransformation)

  /*

* Start time: 2023-04-25T20:22:03.987Z UTC
* Execution time: 208.970 ms
*/
var P = new MorphologicalTransformation;
P.operator = MorphologicalTransformation.prototype.Dilation;
P.interlacingDistance = 1;
P.lowThreshold = 0.000000;
P.highThreshold = 0.000000;
P.numberOfIterations = 1;
P.amount = 1.00;
P.selectionPoint = 0.50;
P.structureName = "";
P.structureSize = 5;
P.structureWayTable = [ // mask
[[
0x00,0x00,0x01,0x00,0x00,
0x00,0x01,0x01,0x01,0x00,
0x01,0x01,0x01,0x01,0x01,
0x00,0x01,0x01,0x01,0x00,
0x00,0x00,0x01,0x00,0x00
]]
];

• Convolution on starmask

• var P = new Convolution;

P.mode = Convolution.prototype.Parametric;
P.sigma = 5.30;
P.shape = 2.00;
P.aspectRatio = 1.00;
P.rotationAngle = 0.00;
P.filterSource = "";
P.rescaleHighPass = false;
P.viewId = "";


• Invert L view
• Pixelmath Inverted L view with Starmask in new file
• Invert this new file and add it as mask to the RGB image.
• This should give you a mask where ONLY the galaxy is not protected.


• Unsharp mask on the image with inverted mask on
• NoiseXTerminator

Mina m.b.:

I'll hopefully make this simple.

Yes, more is possible....  

Generally speaking OSC data takes a huge hit in the red and blue channels due to efficiency passing through the Bayer matrix.   Because most Bayer matrices are RGGB, you are going to get a ton of green long before the red and the blue come in.  (this is why mono is much more preferred for astrophotography).   

What does this translate to?     Noise.   Lack of signal.  

Compared to someone using a mono camera and gathering the same amount of integration time, you're only gathering ~30%.         Your image needs more integration time to fully pop.    If you look at your linear stack in Pixinsight and cycle through the red, green, and blue channels individually you will see exactly what I'm talking about.  


Also, you're using a 183, which unfortunately is not nearly as efficient as today's cameras.     

Judging by your final image, I think you still had some images with clouds in them.   

Overall, I think your image looks just fine for the quality of gear you have and the integration time involved.

Oh, and reading through the comments about aperture size....   I think it has less to do with size, and more the quality of the optics.

It's certainly a possibility I have clouds - at some point, the sky turned fully cloudy, I rejected the worst images according to APP's rating mechanism anyways, but I didn't look at them all individually, to be honest... Now I'm curious what makes you think that, how you spotted that. I realised while stacking that some pictures good a worse quality ranking than others - about a third of the best value - althought, the FWHM wasn't bad, and the guiding / seeing generally constant, so you might be on to something - I might want to go through the subs individually.

Regarding the Mono vs OSC: thanks for explaining, makes sense. I didn't know it was only 30%, that's... rough. I'm eyeing going mono eventually anyways, I can't justify it now, as I want to upgrade my scope first, but next is probably the camera, maybe a 294MM if the 183MM isn't efficient enough... not the biggest fan of the square sensor of the 533MM personally, sadly.

Well, I sometimes feel like I'm hitting the limits of my imaging train, which is starting to annoy me - of course, my post processing isn't perfect, but when the S/R is right, the final result looks better, eg my M31, or the pacman nebula...
The doublet I use has a lot of CA in my opinion, if I don't realign the channels, I have purple fringes on a lot of stars when I pixel peep. After all, it is a 550€ chinese doublet...

Usually any time the stars show up with a fuzz around them, it's either clouds or you dewed up.    Those aren't halos IMO.   

Use Blink process in pixinsight and you can speed through each sub and watch how your seeing improved or worsened throughout the night.   You should be able to see when it got cloudy and reject better.   

This hobby is a money pit.    I'd stay away from the 294MM, it has its own subtle issues. Not that it's a bad camera tho, and those tiny pixels can be nice at times... 533/2600/6200 all have great performance and no amp glow.  

Sounds like you'll end up just like me.  I started this hobby last August with a nearly identical setup.   Within about a month I went full ham.

Something funky definitely happened in the processing to get the weird halos. It is by no means an issue with the equipment (which is great).

As far as "Is there more possible?" Why yes, yes there is. The 183MC through an 80mm aperture at F/5.6(?) is going to have a tough time collecting photons. It is by no means a bad combination but you are certainly starved for light. Assuming you have unlimited clear skies I would of tried for like 8-15 hours with that setup. However, if weather only permitted you to gather what you have now, you can still get a serviceable image out of that.

When it comes to the actual processing, sometimes less is more. I appreciate that you shared the final stack and I gave it a go:

andrea tasselli:
So here we go...