I like my NGC 4565 Needle Galaxy, but I don't like my stars. Any recommendations?

[BrendanC]

Hi all,

I recently took this:

Now, I quite like how the galaxy turned out, but the stars leave a lot to be desired. They seem a bit noisy, bloated and ill-defined. I'm not sure what's wrong. Am I overexposing? Is it noise? Have I overdone the processing? If so, which bit? Are my calibration files crap? Or is it just a basic limitation of my kit? Any/all comments/suggestions/recommendations/ideas/thoughts/opinions/insights/questions welcome.

Here are the details:
• 4:42 hours of integration at ISO 800 from 141x120s subs unguided (240s subs were planned but guiding wasn't working)
• Bortle 4 sky, 45 mins of moon at 6% phase
• Calibration: 25 flats, 25 dark flats, 50 darks
• Hardware: Sky-Watcher 130PDS scope (F5), Sky-Watcher NEQ6 mount, Canon EOS1000D astro-modded DSLR camera with Sky-Watcher 0.9x coma corrector
• Software: polar alignment with SharpCap Pro, capture with Astrophotography Tool (APT), stacking with Deep Sky Stacker (DSS), post-processing with StarTools, Affinity Photo and Topaz Denoise AI

I didn't bin the image. I nornally 2xbin in StarTools to increase SNR, but sometimes, if I want the central object to be larger, I don't. So perhaps that has something to do with it too? If it helps, here's the larger, binned image, in which the stars are a bit better (I think):

Or are they ok and should I stop worrying?

Thanks, Brendan

[teoria_del_big_bang]

That's a great image.

I don't see a lot wrong with the stars, maybe a tad  yellow but that's all, and I am not really experienced enough to give a proper analysis but just looks a fantastic image to me.

Steve

[tomato]

I’ll watch this thread with interest as I have the same issue, although I image galaxies at 0.94 arcsec per pixel, which doesn’t make for tight stars.

Still a very good Needle BTW.

[BrendanC]

Thanks. Although nobody seems to think there's a problem!

[alan4908]

23 hours ago, BrendanC said:

Now, I quite like how the galaxy turned out, but the stars leave a lot to be desired. They seem a bit noisy, bloated and ill-defined. I'm not sure what's wrong. Am I overexposing? Is it noise? Have I overdone the processing? If so, which bit? Are my calibration files crap? Or is it just a basic limitation of my kit? Any/all comments/suggestions/recommendations/ideas/thoughts/opinions/insights/questions welcome.

If you zoom in to a high degree on the unbinned (first) image you can see that all the stars suffer a severe elongation. You can also see that the noise is quite high and that the RGB colours and not quite correct. 

I suspect that the non-round stars are due to the fact that your guiding was not working and so the lack of roundness is simply a manifestation of your mounts tracking inaccuracy.  If you look at the binned image, where you have decreased the overall resolution of the image, they appear more round. You can also fix this effect to a large degree in post processing -  I'd suggest Photoshop -  take the image, duplicate the layer. Set the top layer to Darken and then apply the offset filter to the top layer (this moves the top layer by fractions of a pixel with respect to the bottom layer), until you get rounder stars.

To improve your RGB star colours I'd suggest you separate the result into an RGB and a luminescence image and process these separately. This relies on the fact that almost all the detail comes from the luminescence rather than the RGB image. So, take the RGB image and blur it until all the stars look quite blurry. You might also want to increase the colour saturation slightly.  Then recombine if with the lum image. 

On the noise front - I'd make the background less dark - in Photoshop I'd suggest a background level of between 18 and 24, your image is currently around 3 to 5.  I then suggest you look at the image and see where the main sources are, in general I'd always suggest applying any noise reduction scheme in conjunction with an object mask so that it is targeted where it is most needed.

Hope this helps.

Alan

[BrendanC]

Thank you!

I think you're right about the elongation of the stars. I was forced to use 120s subs because guiding wasn't working at all (turned out I'd moved the guide camera and PHD2 didn't like it until I recalibrated the next day), and I thought I could push the mount to that time without guiding. Looks like I pushed it too far. It's an old(ish) NEQ6 but has the belt mod. Maybe I should get it serviced or something sometime.

The star colours might be fixed from now on because I discovered how to use the Max RGB feature in StarTools (which I use for most of my post-processing, I generally just use Photoshop and Affinity for additional tweaks). I discovered that I had a lot of green in one image, which might mean I can go back and fix the others sometime.

Interesting idea about processing RGB and Lum separately. I used to do that in Photoshop to stretch each channel but not since using StarTools. I don't even know how to do that in ST, or even whether it's possible from a DSLR image. But I'll have a look.

For the noise, I use Topaz AI and it works well, but recently I've noticed more noise than I'd like. So, last night, I ran a test on M64, in which I did equal times of 120s, 180s and 240s exposures. I'm actually processing them all now, so let's see which one yields the best results.

Really appreciate the feedback. Thanks again.

Cheers, Brendan