StarTools

Hi all,

inspired by Carles’ nice wide field image of NGC 6914 I decided to try this object too – a bit more more of a close-up at 1000 mm focal length:


Conditions were quite nice, but I’m still struggling with my self-build dew cap, which caused some harsh gradients, probably due to stray light falling in. So I had to discard some hours

And again visible problems with this strange kind of halo. The bright blue star in the upper left corner has no reflection nebula around it for sure. And it might cause a slightly blurred overall impression.

This is the first time I have used the NB Accent module to process a nebula (only one or two galaxies so far). Wow – this works so well! Really awesome. I dialled in everything - a real eye-burner for my standards Processing with ST was pretty straight forward. I used the Detector Gamma of OptiDev to bring in some more brightness. SVDecon only with 5 PSF iterations as more added too much noise, but it was helpful for the stars, anyway. No HDR, no Sharp, no Shrink. I followed Stefan's advice and first ran NBAccent and then afterwards the SuperStructure module.

Please see here for technical details:

https://www.astrobin.com/xc0kyo/

Comments welcome, as always.

Best regards, Dietmar.

Hi Dietmar,

interesting to see that the target attracts quite some imagers. Mike and myself also already gave it a try But maybe not surprising since it contains emission and reflection nebulae as well as some dark dust lanes. What else can you wish for?

Really good close up of the field which was rather wide field in the images here in the gallery. Lots of nebulosity and also quite some signal in the darker parts. I missed that

Two remarks, basically personal taste: I probably would have stretched a little bit less. This would maybe increase contrast a bit and counteract the slightly blurred appearance you mentioned. Or increase locality in the Contrast module (after applying NBAccent).

With regard to color and looking at the histogram on Astrobin - maybe green is reduced a bit too much, resulting in a slightly purple cast. Color is of course always a matter of debate and personal taste By the way, I wondered why your histogram has so many colors and mine basically only three Is your image in the RGB color space or did you convert to YMCK? But maybe I am totally off here...

Anyway, really good work. I always enjoy looking at NBAccent images

Regards
Stefan

Hi Stefan,

thanks - and thanks for your comprehensive reply

Stefan B wrote: ↑Tue Aug 20, 2024 7:26 am also quite some signal in the darker parts.

Yes, I too was surprised what came out. But probably this was the reason for me that I stretched it so (too?) much.

Stefan B wrote: ↑Tue Aug 20, 2024 7:26 am Or increase locality in the Contrast module (after applying NBAccent).

I will try that - I already played a bit with contrast changes after processing in StarTools and indeed that may help a bit.

Stefan B wrote: ↑Tue Aug 20, 2024 7:26 am maybe green is reduced a bit too much, resulting in a slightly purple cast.

Not sure - maybe the reason is that I used Balmer Series response, but it was Pure Red in your case? I will check both options.

Stefan B wrote: ↑Tue Aug 20, 2024 7:26 am By the way, I wondered why your histogram has so many colors and mine basically only three Is your image in the RGB color space or did you convert to YMCK? But maybe I am totally off here...

No idea what you are talking about. I'm not feeling guilty in any way I'm using darktable (photography software) to export to JPEG format. I don't see any difference compared to the original TIFF produced by StarTools.

Best regards, Dietmar.

Hi Dietmar,

decay wrote: ↑Tue Aug 20, 2024 4:36 pm Stefan B wrote: ↑Tue Aug 20, 2024 9:26 am
maybe green is reduced a bit too much, resulting in a slightly purple cast.

Not sure - maybe the reason is that I used Balmer Series response, but it was Pure Red in your case? I will check both options.

That may also be the case. But looking at your histogram on ABin


it could be that the channel alignment is a bit off. Green starts first, then comes blue, then red. From most of my quite red NBAccent images I am used to see that green and blue align well in general and that red has more of the brighter pixels of course but that curve of the red channel starts at the same point of the histogram. This for example is the histogram of my brutally red Pelican image:


You can see that blue and green almost overlap and red is totally different. But all three channels start almost at the exact same spot on the left. The histogram of my NGC 6914 is actually not a good example:


Here the red channel is probably pushed a bit too much. I sometimes see this taken to the extreme when people post images where there seems to be Ha everywhere and they get comments like "Woah, you captured an insane amount of Ha signal" but actually they just stretched red so much that they tainted their neutral background in red.

Anyway, I hope you know what I am aiming at. I am no expert at all in signal processing or interpreting histograms but that's how I (try to) read them.

decay wrote: ↑Tue Aug 20, 2024 4:36 pm By the way, I wondered why your histogram has so many colors and mine basically only three Is your image in the RGB color space or did you convert to YMCK? But maybe I am totally off here...

No idea what you are talking about. I'm not feeling guilty in any way I'm using darktable (photography software) to export to JPEG format. I don't see any difference compared to the original TIFF produced by StarTools.

So, well, that's embarrasing for me. I meant the CMYK color model (https://en.wikipedia.org/wiki/CMYK_color_model) but obviously was thinking of the song by the Village People Anyway... you can choose CMYK for example in Photoshop instead of the default RGB. I was thinking that this maybe changes the histogram display or the curves on ABin. For comparison your histogram and mine:


Yours is much more...don't know how to say...divers? Interesting? Mine has just three almost Gaussian distributions (well, yes, not the red channel; blame all that Ha for that).

Here's Mike's NGC6914 histogram:


Just wondering why yours looks so different. And I don't want to indicate that yours is wrong. Maybe mine is. Or both. Or none of them Maybe it's really just Balmer vs pure red as you said.



Regards
Stefan

This is by the way the histogram of my Pelican image where I used the Balmer series option and cyan for OIII in NBAccent:


So basically a blue and green overlap like in my 'pure red' NBAccent images. So Balmer series doesn't result necessarily in a drastically different histogram than pure red.

I recently tried to cheat in my Pelican image since I still didn't like the colors. I used the selective color adjustment tool in Photoshop which I haven't done in years (had to google where to find it). But interestingly what you get is this:


Here's the image to it: https://www.astrobin.com/w0fan0/D/ The green and blue overlap gets obviously less.

Regards
Stefan

Hi Stefan (@Stefan B),

so much input - I will need some time to think about and to check all this.

This is what I already can say: The left parts of the histograms where usually green and blue go together are the darkest areas of the images. Probably a colour cast is simply not visible in this areas because it's too dark to see? So the interesting action takes place more towards the middle of the X axis I guess. And indeed here blue is much stronger than green in my case and this causes the purple cast. So yes, maybe green is reduced too much. I will check that.

Not sure if this also is the reason why my histogram looks more 'divers'. I think I will try to reduce green and then maybe it will be not be 'divers' any longer? We will see.

As for the YMCA colour model which has it's technical founded origin in physical printing processes, I believe it's use in software algorithms is just another way / model of representing or working with colours. And in the end everything gets mapped onto RGB values, anyway. However, I'm quite sure that my JPEG export is straight forward and there doesn't seem to be any difference to the original. So I don't think this is the reason for my abnormal appearing histogram

Dietmar.

P.S.: Pelican revision D is the best so far - cheated or not

Hi Dietmar,

decay wrote: ↑Wed Aug 21, 2024 6:31 pm The left parts of the histograms where usually green and blue go together are the darkest areas of the images. Probably a colour cast is simply not visible in this areas because it's too dark to see?

I think the overlapping regions can be seen. Here's the histogram of an image of the Double Cluster (https://www.astrobin.com/050lne/):


Since stars emit continously blue, green and red all parts mostly overlap and the curves are in a region which also corresponds to the overlap in our nebula images.

decay wrote: ↑Wed Aug 21, 2024 6:31 pm So the interesting action takes place more towards the middle of the X axis I guess. And indeed here blue is much stronger than green in my case and this causes the purple cast. So yes, maybe green is reduced too much. I will check that.

Will be interesting what you'll find out.

decay wrote: ↑Wed Aug 21, 2024 6:31 pm As for the YMCA colour model which has it's technical founded origin in physical printing processes, I believe it's use in software algorithms is just another way / model of representing or working with colours. And in the end everything gets mapped onto RGB values, anyway. However, I'm quite sure that my JPEG export is straight forward and there doesn't seem to be any difference to the original. So I don't think this is the reason for my abnormal appearing histogram

You're probably right about that...

decay wrote: ↑Wed Aug 21, 2024 6:31 pm P.S.: Pelican revision D is the best so far - cheated or not

At the moment I agree

Regards
Stefan

Hi Stefan,

Stefan B wrote: ↑Tue Aug 20, 2024 7:26 am maybe green is reduced a bit too much, resulting in a slightly purple cast

I followed your advice and and dialed in some more green. Indeed this seems to help Thank you for pointing out!
Additionally I switched from Balmer Series response to Pure Red. Now it looks more like most of the images we can see on AB for example.

Stefan B wrote: ↑Tue Aug 20, 2024 5:51 pm Here the red channel is probably pushed a bit too much. I sometimes see this taken to the extreme when people post images where there seems to be Ha everywhere and they get comments like "Woah, you captured an insane amount of Ha signal" but actually they just stretched red so much that they tainted their neutral background in red.

True. I think I did the same with NB Accent. This time I reduced the impact of NB Accent a bit and I think this looks more appealing. What do you think?

Here's the histogram:


So the left side now looks more like you would expect. But nonetheless it's still pretty 'divers', isn't it? Well, I dunno ...

Best regards, Dietmar.

Hi Dietmar,
This rendition is really good and colour with Ha accent is excellent. I don’t know this celestial object being from down under but it seems like you’ve pretty much nailed it.
I had colour issues with my recent first posted M20 with Ha accent and all I did was increase the Green bias and use the Ha accent Galaxy preset ( red) instead of Nebula preset ( Balmer ). The pinks and purples changed to pinkish red and purples to blue.
Kudos to Ivo for the Ha accent feature, it can really make your images pop. I must use on my next galaxy image.

Well done !!

Martin

Hi Dietmar,

histogram still looks a bit funnier than most others but the image looks really good in my eyes. And that's what matters more than a pretty histogram, right The pure red does a better job than the Balmer simulation here in my opinion.

Pretty good contrast, too. Did you adjust that compared to your previous rendition? Shadows appear a bit less stretched

Regards
Stefan