Pacman nebula narrowband

[poogle]

Hi

This is my first go at combining Ha, S2 and O3 into a color image. Narrowband is the way to go for me since I live next to an airport in Stockholm, Sweden. The image consist of around 2 hrs of 10 minute subs in each channel, combining for 6.5 hrs.

I'm not very fond of processing and I'm sure there's more to be found in there somewhere

Got some strange artifacts during stacking that I've never seen before, see attached image. Anyone have any suggestions on why?

Thanks for looking

/Patrik

[poogle]

Forgot to do the magic SCNR

[glowingturnip]

nice image, and yes, there probably is more to be got out of it with processing, especially if you've got 6+ hours integration time ?

Just wondering how you combined the Ha, Sii and Oiii ?  The usual SHO combination would have a bluish middle and orange edges, whereas yours looks more like an RGB image ?

Not sure about the cause of the black specks, but if you make sure you dither between subs, and if you're using Pixinsight, run CosmeticCorrection targeting the dark pixels on each individual sub prior to stacking and that should get rid of most of them.   

[poogle]

1 hour ago, glowingturnip said:

nice image, and yes, there probably is more to be got out of it with processing, especially if you've got 6+ hours integration time ?

Just wondering how you combined the Ha, Sii and Oiii ?  The usual SHO combination would have a bluish middle and orange edges, whereas yours looks more like an RGB image ?

Not sure about the cause of the black specks, but if you make sure you dither between subs, and if you're using Pixinsight, run CosmeticCorrection targeting the dark pixels on each individual sub prior to stacking and that should get rid of most of them.   

Thanks for the input, I'll give Cosmetic Correction a go. I think the specks originate from the stacking though, since I didn't see them in the subs at a first glance, but might be mistaken 

You're right about the colors, Ha is mapped to red, O3 blue and S2 green. I've never tried the Hubble palette, but will give that a try as well!

[poogle]

Tried a quick hubble edit, but I think I need alot more S2 and O3 data for it to work. Found it difficult to control the green in the background (probably due to lack of PixInsight skills  )...Will give it another go tomorrow.

CosmeticCorrection didn't do the trick for me, I still ended up with the black specks in the stacked Ha image  I've tried all stacking methods in DSS and can't see any black specks in the single exposures, so I'm clueless. Will try to learn stacking with PI

/Patrik

[Jbro1985]

By no means perfect, but this is a similar integration time in SHO from about a month ago (also LP skies, next to Heathrow!).  SHO brings out a lot more detail.  Have a look at light vortex online or YouTube search pixinsight narrowband. 

Essentially, I've found the best results to be combination through pixelmath (no formulae just mapping (don't know why this works better than standard combination but seems to)), SCNR to take some of the green out and then careful use of colour masks with the curves tool.  Just to note, this step can introduce a bit of noise so need to be careful not to push too hard.

I too suffer from the black spots but use CC in pixinisght to kill them off.  Out of interest, what camera are you using?

[Jbro1985]

Nb.  If SCNR struggles with the green you might try applying the stretch (I used ArcSin on the above) and specifically use the ColorMask tool to generate a green mask, apply the mask and change the colour balance on the green stuff manually via the split colour channels on curves as an alternative workflow.  

[poogle]

Thats amazing Jason. So much detail and a nice black background with great contrast!

Here's my quick Hubble attempt using PixelMath, which is nowhere near your image. During stacking I reused flat frames from a previous session (I know I know... ? ), which resulted in a quite nasty gradient that was hard to cope with, hence the massive cropping and halos around the brightest stars. I've tried the tutorials on LightVortex, but my attention span might be a bit short for narrowband image processing. I just want to stack the images in DSS and give it a quick 1 hour process and move on to the next target  But looking at your image it sure is worth it.

I image with a SXVR-H18. I think the tracking, focus and integration time is good enough to produce much better end results.

/Patrik

[glowingturnip]

I'd say firstly, ditch DSS and learn Pixinsight stacking, you should find it far superior (I'll probably get contradicted saying that !) - try Harry's tutorial for learning it - https://www.harrysastroshed.com/pixinsight/pixinsight video html/pixinsighthomenewbie.html

So in your original image, you've got a mostly red nebula since as you say the Oiii and Sii signals are much weaker than the Ha, so they're not actually contributing to your nebula at all there, they're just giving you better coloured stars.

In your second image, you've got a lot of green noise because the Ha channel has now been over-stretched in your attempts to bring out the other channels.  There's some Oiii showing now, but still hardly any Sii.  What they really need is that each channel has its own stretch prior to being combined together, so that they're all able to contribute evenly - that can lead to issues with stars (the infamous magenta halos).  A method I like is a tonemap method, detailed here, though it it rather complicated - http://www.arciereceleste.it/articoli/translations/75-narrowband-color-composition-eng

[Jbro1985]

Once you're used to it, it's no more complex than lrgb processing really and you do the have to deal with LP gradients.  

As above, get involved in the PI stacking and calibration, it's well worth it.  

I would say though if you want good clean images you do have to submit yourself to the processing.  For instance ABE in PI is quicker than DBE, but most of the time won't yield as good results.  Same can be said of using deconvolution and making sure you use dynamic psf first and get a star mask in place for local deringing support. 

Like acquisition, there's a time investment involved.  If you're not so fussed and enjoy the rest more than processing, use a simpler program than PI as you'll probably get the same results.  You might even find an hour on a different software yields a better result than PI, but if time is not in question and you know how to use it, PI will likely yield THE BEST result more times than not.

[wimvb]

Very nice image. The black specks can be a result from too severe clipping. Either during image calibration, integration (sigma clipping), or stretching.

[tooth_dr]

1 hour ago, wimvb said:

Very nice image. The black specs can be a result from too severe clipping. Either during image calibration, integration (sigma clipping), or stretching.

I was stacking data from two cameras, both with KAF-8300 sensor, and got the same specks as above.  I have come to the conclusion that the integration was at fault.

[poogle]

Thanks alot for the peptalk Jason ? I gave it another try, with both stacking and processing in PI using various tutorials.

Still lightyears away from your image, but at least I'm getting some better results! I have a lot of learning to do...

I think the black specks came from clipping during integration as suggested Wim, playing with the values gave a better end result.

 

[wimvb]

Looks great!

With your last image, you can try scnr at 90 % to tone down the green a bit. The Hubble palette contains some green, but it may be more appealing if the other colours also come through. For the modified Hubble palette, use scnr at 100 %. From here on it's all about playing with settings to try different variations.

[poogle]

Wim: just wow! Thanks for the tip! The amount of skill on this site is amazing!

Sorry for spamming, but gotta post the edit after SCNR  Looking through the image progress in this post is kinda comic, lol

 

[wimvb]

I took the liberty of downloading your jpeg and had a quick go at it:

hubble palette with scnr applied through filter (green channel of the original)

Modified hubble palette (scnr green at full strength):

Average of the above:

Btw, what airport are you near; Arlanda or Bromma? Both are pretty awful, as far as light pollution is concerned.

[poogle]

Ah, I think I misunderstood you, I just ran a 90% scnr on the original, but averaging them out like you did looks better. Will have another try tomorrow, starting from the beginning. I'm getting hooked by the processing now 

It's Bromma Airport, I'm glad I didn't get any airplane trails in any of my subs

[wimvb]

2 hours ago, poogle said:

Ah, I think I misunderstood you, I just ran a 90% scnr on the original

No, you understood me allright; I just tried a variation. Since I tried to remove green from the image, but didn't want to affect anything else, the most obvious mask was the green channel from the image. I boosted this somewhat (histogram transformation) and applied it before scnr at 100%. Then did scnr without a mask on a clone, and combined the two with pixelmath:

mean(image, image_clone)

Also did a little noise reduction (TGVDenoise on Luminance, and MMT on Chrominance). Results will be better on the original than on a compressed jpeg copy.

[glowingturnip]

15 hours ago, poogle said:

Wim: just wow! Thanks for the tip! The amount of skill on this site is amazing!

Sorry for spamming, but gotta post the edit after SCNR  Looking through the image progress in this post is kinda comic, lol

 

 

that's looking more like it !  ?