Flying Bat, Barnard150, OU4 and dust.

[ollypenrice]

Paul Kummer and I have extended our 2 panel Bat to 3 panels to include the dusty Barnard 150. This was a good suggestion of Gorann's, and SGL member George Sinanis kindly supplied nice OIII data to bring in Outers 4, the Squid.   The spirit of this image involves featuring the dust so there is no Ha here, just OSC and George's OIII, without which the Squid is invisible.

It's always pleasing to mention the name of the great and honorable Edward Emmerson Barnard, who was both the last of the great visual observers and the first of the great astrophotographers.

Olly

[vlaiv]

I'm getting distinct - "over sharpening on the stars" vibe from the image.

 

[ollypenrice]

44 minutes ago, vlaiv said:

I'm getting distinct - "over sharpening on the stars" vibe from the image.

 

I actually gave them a Gaussian blur of 0.6 in Ps and applied the star layer at slightly less than full opacity to soften them. Perhaps I should gave given them more blur...

Olly

[vlaiv]

3 minutes ago, ollypenrice said:

I actually gave them a Gaussian blur of 0.6 in Ps and applied the star layer at slightly less than full opacity to soften them. Perhaps I should gave given them more blur...

Olly

I think that is actually down to blend mod.

Try "lighten" for example.

Here - look at this part:

Those larger stars look like they have dark halo around them. This is due to blending, but if you over sharpen things - you get the same effect.

[ollypenrice]

1 hour ago, vlaiv said:

I think that is actually down to blend mod.

Try "lighten" for example.

Here - look at this part:

Those larger stars look like they have dark halo around them. This is due to blending, but if you over sharpen things - you get the same effect.

They do, I know, and I've mitigated that effect to some extent by lifting the shadows around them. I used to use blend mode Lighten to re-apply stars but now greatly prefer Screen. There's a third way which I haven't yet tried.

I do think that the re-application of stars is an area to work on but, overall, star removal takes us way deeper into the nebulosity and I'm happy with what I'm doing as a 'method in progress.'

Olly

[Stuart1971]

3 hours ago, vlaiv said:

I'm getting distinct - "over sharpening on the stars" vibe from the image.

 

I didn’t want to say anything, but I also thought the stars looked odd in places…but then what do I know…..🤔

Edited September 14, 2023 by Stuart1971

[vlaiv]

9 minutes ago, Stuart1971 said:

I didn’t want to say anything, but I also thought the stats looked odd in places…

the Emperor is not wearing anything at all!

[gorann]

Nice framing Olly and thanks for following my suggestion of a second panel😃 Now you just have to add one more to the right to get a very pretty galaxy in there:

https://www.astrobin.com/lnhy4h/

[simmo39]

Stunning image to my eyes, v 3d.

[ollypenrice]

OK, I've softened the stars a little - though I'll continue to experiment with ways of recombining them with a starless layer.

Thanks for the critiques.

Olly

[Elp]

I assume you've already done the star minimisation technique? Whilst doing it you can also gaussian blur very slightly which softens the star edges.

[Laurin Dave]

Very nice Olly..  with my Samyang setup I’ve found that stars look best without BXT, removed after a light stretch, PS colour enhanced using Noel’s action, blurred, then slightly sharpened with UnSharpMask..   if needed remove blue halos with Noels action and red halos by inverting and running HLVG (or scnr) then added as multiple Screen layers to taste…. the stars get more stretched the more screen layers you add..  and you can just erase brighter stars from one or more screen layers to reduce them ..  

ps I always take them stars out in Pixinsight after the light stretch with ‘unscreen’ ticked in SXT

Edited September 15, 2023 by Laurin Dave

[vlaiv]

1 hour ago, ollypenrice said:

though I'll continue to experiment with ways of recombining them with a starless layer.

If you don't mind me asking - how do you extract stars in the first place?

Do you create stars only image at all or do you work with integral data for stars?

[ollypenrice]

5 hours ago, Elp said:

I assume you've already done the star minimisation technique? Whilst doing it you can also gaussian blur very slightly which softens the star edges.

I have total control of star size when re-applying them using blend mode screen and, yes, I gave them a Gaussian blur but maybe not enough. Now increased.

5 hours ago, vlaiv said:

If you don't mind me asking - how do you extract stars in the first place?

Do you create stars only image at all or do you work with integral data for stars?

Run Star XT then here's the workflow in Ps.

StarXterminator workflow in Photoshop.

1 Stretch standard image as usual to about 60 to 80% of final stretch. Simple stretch, nothing detailed. Save as Proc 1.

2 Run star Xterminator and save the starless image as Starless.

3 Continue to process Starless. Cosmetic repair of artifacts, harder stretch, contrast enhancement, noise reduction, sharpening, colour etc. Save. Select and copy.

4 Use Open Recent to re-open Proc 1 and paste Starless as a top layer.

5 Invert both layers.

6 Top layer active, change blend mode to divide.

7 Stamp down. (Alt Ctrl E)

8 Top layer active, Invert.

9 Flatten Image. (I seem to have to do this from the toolbar Layers dropdown because CtrlE doesn’t work.)

10 Save as Stars.

11 Select Copy

12 Paste onto Starless.

13 Blend mode to Screen.

 

Actions 5 to 9 inclusive can be recorded as a single action.

14 The stars can be reduced simply by lowering the mid point in Levels. Small stars can look too hard and can benefit from the simple contrast tool to reduce contrast. Large stars with halo or bloat benefit from an increase in contrast. Other possibilities include Gaussian blur or a reduction in top layer opacity by a tiny amount. Stars which look ‘stuck on’ over nebulosity can be made to settle into the image by means of a dab with the burn tool on the bottom layer, just underneath them.

Olly

Edited September 15, 2023 by ollypenrice
Clarification

[Elp]

You can also use the minimise filter to reduce star size, but only by one or two pixels otherwise smaller stars tend to disappear. Your stars have to be selected first with a highlight selection with the edges feathered. Gaussian can be applied at the same stage whilst the stars are still selected, again only by a fraction as you'll likely know.

Edited September 15, 2023 by Elp

[vlaiv]

13 minutes ago, ollypenrice said:

Run Star XT then here's the workflow in Ps.

I'm not sure how keen are you on changing the workflow that you are comfortable with, but may I suggest something a bit different.

I'll use simple explanation for what is going on rather than exact technical terms, but hopefully that will be enough to understand. This workflow is for luminance only - it does not preserve star color in cores as it leads to clipping, so you'll need to restore those with some mask or something.

1. Do levels / linear stretch until you start seeing nebulosity showing up. This step is essentially just linear transform bringing white point down.

2. Do gamma adjustment (or middle slider in levels) - to a controlled value (say 2.2 - 2.4. Not sure about PS but Gimp has feature where you can just enter the number).

3. Adjust black point to your liking.

This is minimal stretch that will let star removal algorithms work - and it is actually reversible if there is need.

Make a copy of above image and run starless routine on it - this will produce starless image.

Now you need to do pixel math of sorts and simply take original copy and subtract starless image. This will give you stars only image. You will separately stretch that image just to a point where you are happy with stars. You might as well leave them at that level.

Process starless image as you would and then layer stars only on top of that with brighten mode (should be really add mode, but brighten will work as well).

[ollypenrice]

2 hours ago, vlaiv said:

I'm not sure how keen are you on changing the workflow that you are comfortable with, but may I suggest something a bit different.

I'll use simple explanation for what is going on rather than exact technical terms, but hopefully that will be enough to understand. This workflow is for luminance only - it does not preserve star color in cores as it leads to clipping, so you'll need to restore those with some mask or something.

1. Do levels / linear stretch until you start seeing nebulosity showing up. This step is essentially just linear transform bringing white point down.

2. Do gamma adjustment (or middle slider in levels) - to a controlled value (say 2.2 - 2.4. Not sure about PS but Gimp has feature where you can just enter the number).

3. Adjust black point to your liking.

This is minimal stretch that will let star removal algorithms work - and it is actually reversible if there is need.

Make a copy of above image and run starless routine on it - this will produce starless image.

Now you need to do pixel math of sorts and simply take original copy and subtract starless image. This will give you stars only image. You will separately stretch that image just to a point where you are happy with stars. You might as well leave them at that level.

Process starless image as you would and then layer stars only on top of that with brighten mode (should be really add mode, but brighten will work as well).

I'm working with OSC data. I realize that I could extract a luminance layer and process that as you suggest, then process the OSC data as I would have processed an RGB layer, the objectives being rather different in processing L and RGB.

However, when StarX appeared, I (and many others) extracted the stars by subtracting the starless from the starry as you suggest. I never found that this worked as well as the routine I follow now.

I do like my present routine though it involves making some calls which vary from image to image and sometimes requires small stars to be processed differently from large ones.

3 hours ago, Elp said:

You can also use the minimise filter to reduce star size, but only by one or two pixels otherwise smaller stars tend to disappear. Your stars have to be selected first with a highlight selection with the edges feathered. Gaussian can be applied at the same stage whilst the stars are still selected, again only by a fraction as you'll likely know.

There is absolutely no need for this. The minimize filter, so far as I can see, belongs to the older processing techniques in which stars were not extracted. When they have been extracted, and the starless image has been given a very hard stretch without them, the extracted 'stars-only' can be replaced as a top layer. Using the grey point slider you can take them from entirely invisible, through tiny to small, medium and then large. What would I want to do with the minimize filter? The grey point slider gives me total control of star size.

Olly

[david_taurus83]

Another nice transparent Squid, love it!

I know some people put a colossal amount of hours into this object with OIII filter to force it out in processing (and I appreciate the time and effort) but I much prefer to see it faint and see through like this and another a few days ago.

[ollypenrice]

Agreed. Our priority in this rendition was the dust rather than the gas.

Olly