View Full Version here: : Trying a more refined starless image...
Benjamin
19-04-2020, 01:39 AM
Had a go at removing the stars in this version of IC 2944 in PixInsight using MMT and star masks. Lots of control of the process but then lots of umming and ahhing... I usually remove stars in quite a rough way as I add them back in later but for this I tried to maintain some detail.
Still a HOO image. Have some SII and RGB but yet to add this to the mix.
Astrobin: https://www.astrobin.com/u8txvi/
Andy01
19-04-2020, 07:56 AM
Sorry Ben but all the nice detail has been lost in your technique used here. :sadeyes:
Your original has much more, but this starless version, while dreamy and painterly, has lost all of the interesting nebulosity so the bok globules, misty mountain & coal miner are hard to make out. :question:
At risk of sounding like a complete tosser, if you compare it to this one (https://www.astrobin.com/full/406335/B/), you'll see what I mean. :)
Starnet++ is awesome software and can produce great results, but I have yet to see it used very well with the pixinsight module. :shrug:
The colour palette however, is lovely, so have another go mate! :thumbsup:
Benjamin
19-04-2020, 09:10 AM
Thanks so much for your response Andy and that amazing image. I’ve made some adjustments now. It’s a bit “post-process”’ in that I’ve tried to sharpen what I have than redo anything. Can I ask about the process you use to sharpen the features (in general terms perhaps) or what steps you take to keep detail in the Star removal process? I think my misty mountain is probably limited by the star removal process now so would require a redo although the next plan is to add SII which could help definition?
Astrobin: https://www.astrobin.com/full/u8txvi/B/?nc=user
multiweb
19-04-2020, 09:46 AM
Looks great Ben. :thumbsup: Can't tell the difference. :question:
Benjamin
19-04-2020, 10:03 AM
The two images below are from the core one less sharp I think but maybe the differences are not significant enough? The Astrobin image should be a bit clearer than these of course.
multiweb
19-04-2020, 10:06 AM
Yeah one is a bit sharper than the other but I didn't understand what was wrong with the original. Looked alright to me.
Benjamin
19-04-2020, 10:22 AM
Cheers Marc. I was just following up on Andy’s suggestions and if I could get half way close to his amazing version I’d be a happy soul
multiweb
19-04-2020, 10:30 AM
TBH Andy's a little noiser which might make it appear sharper visually like an apodisation mask, but you have a higher SNR and as detailed so a more natural look. I wouldn't worry about it. It's just a perception.
Andy01
19-04-2020, 02:19 PM
Hi Ben,
Your new version is a vast improvement, but I suspect that re-doing the star removal process from scratch will yield an even better result. :)
I use starnet++ from terminal on my mac. I just let it run, no masking required and it just does it's thing.
Then I layer this starless version over the original in PS and blink between the two to be sure it hasn't introduced any artifacts or odd results.
I'll then mask out any defects and once I'm happy with the starless result, I'll run it through either Topaz sharpen AI and/or Denoise AI to see if can be further improved without introducing any artifacts or tell tale oversharpening etc. :D
Starless images imo can be absolutely stunning, and are popular but they should be starless only, not detail less! :lol:
Hope that helps! :thumbsup:
Benjamin
19-04-2020, 02:46 PM
Thanks Andy. Helps a lot and I think can translate that into PixInsight! I think a redo of the starless Ha image I used for the luminance layer might help. And yes, the starless thing is kind of cool but also easy to do poorly, which is why I’m having a crack. Very wary of noise artefacts which is why I under did the shaprening but it looks like it can take some more perhaps....will have another go with some added SII data I think as this really emphasizes the features.
Phil Hart
19-04-2020, 06:04 PM
That's a beautiful image Ben.
I've just discovered Starnet++ also. You can use it directly as a module in PixInsight. I'd be very interested to hear how its results compare to your manual MMT approach.
My first use leaves quite a lot of artefacts behind but I'm working with broadband and widefield colour image rather than narrowband which I imagine would make it a fair bit easier.
Phil
Nikolas
21-04-2020, 02:26 PM
Looks like an amazing backdrop for a sci fi movie stunning, one of th best starless images I've seen
Benjamin
21-04-2020, 03:23 PM
Thanks Nik and Phil. I've since added a SHO version to my Astrobin gallery. I did a version (not put up anywhere) that tried to go sharper but up close it looked too much like pencil drawing to me. Maybe that's not a bad thing from a distance? Anyway, I might yet get more detail with a reprocess.
Astrobin: https://www.astrobin.com/full/s81kbk/0/?nc=user
marco
21-04-2020, 09:35 PM
Great image Ben, the starless version in particular is very fascinating, the color scheme is pretty good on this one :thumbsup:
Clear Skies
Marco
Benjamin
22-04-2020, 06:27 AM
Thanks Marco. I’ve been playing with colour a lot on the these images but in the end there seems to be a HOO or SHO ideal in my head (right or wrong) that I keep returning to.
multiweb
22-04-2020, 10:14 AM
Beautiful repro. I like the color saturation and the aqua blues. :thumbsup:
Benjamin
22-04-2020, 04:05 PM
Thanks Marc. Without stars you can certainly push the colours a bit
RugbyRene
24-04-2020, 03:17 PM
Excellent image. Has a very 3D look to it.
Rene
gregbradley
24-04-2020, 06:14 PM
It looks just like a fire don't you think?
Greg.
Benjamin
24-04-2020, 11:10 PM
Thanks Rene. When framing it this way Greg I had in mind some kind of steaming cauldron. Chicken soup?!
vBulletin® v3.8.7, Copyright ©2000-2025, vBulletin Solutions, Inc.