fringe killer filter add blue back to central star

Requests for new features and wish-list items.
User avatar
admin
Site Admin
Posts: 3364
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: fringe killer filter add blue back to central star

Post by admin »

alacant wrote:I think it's good to have a go with what you have rather than not having a go at all. There's no time soon when I'm gonna be able to get an APO so for the time being, this is great.
You're absolutely right!
Ivo Jager
StarTools creator and astronomy enthusiast
alacant
Posts: 211
Joined: Mon Jan 25, 2016 7:03 am

Re: fringe killer filter add blue back to central star

Post by alacant »

The above was accomplised by applying decon (in Planetary mode, e.g. no mask used at all) to the blue channel (all image pre-binned to 50%), shrinking the stars using the Magic module and re-merging it with red and green in the LRGB module.
The Develop module was used in the manner described above ("stretch-as-is") after I used AutoDev with a ROI over a star for the general stretch.
The final result (after switching Tracking off) was put through the Fringe Killer to remove any strong/deep blue.
Hi Ivo. Sorry to bring this back again but I've come back after a few weeks and I've lost the workflow. Could you give me a step by step? This is what I've tried, but nada...

TIA and clear skies,
Steve

gimp:
split channels

StarTools:
load blue
bin 50%
autodev roi on a star
dev stretch as is
decon planetary
track off no noise reduction
magic
save

merge with r and g
(rest of my normal workflow here)
User avatar
admin
Site Admin
Posts: 3364
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: fringe killer filter add blue back to central star

Post by admin »

No worries Steve!

This should get you on the right track;

--- Bin
Parameter [Scale] set to [(scale/noise reduction 50.00%)/(400.00%)/(+2.00 bits)]
I binned the image to 50% and saved it.

--- LRGB
Click "blue" to load only the blue channel of the file we just saved. The other channels will be automatically set as well due to "Channel Interpolation". So you end up with a b&w representation of the blue channel.

--- AutoDev or Develop
So we can see what we're doing (we'll be undoing this stretch later on using the "restore" functionality).

--- Deconvolution
There's a few different ways you can try to mitigate the fuzziness of the stellar profile.
You could do it as described earlier, or do it as follows; generate mask automatically.
Parameter [Mask Behavior] set to [De-ring Mask Gaps, Show Result]
Parameter [Radius] set to [3.2 pixels]
Parameter [Iterations] set to [1]
Whatever your technique or values you choose, the goal is to make the stars more pin-point with better defined cores (so this better matches the other channels which also have more pin-point and better defined cores).

Now you can use StarTools' unique "Restore" functionality to restore the data to a state where it is unstretched (so we can merge it with the unstretched red and green channels), but with deconvolution still intact (the Restore functionality is a perk of the Tracking feature of StarTools, allowing you to "time travel" and pick and choose aspects of the image you like - an "undo" feature on steroids). We choose "linear, wiped, deconvolved".

Save the image.

Now use the LRGB module again to load the red and green image from the 50% binned (but otherwise untouched) image, and load the image we just saved as the blue image.

You should now have an image with a (somewhat) better focused blue channel.

From here, you can use process the image including the Luminance contribution trick and Fringe killer as outlined before.

When using the Color module, the Color constancy feature ("Scientific color rendering") may not work very well in the cores due to the presence of ringing artifacts in the blue channel (due to the deconvolution). You can use "Artistic, Detail Aware" to keep StarTools from trying to recover color in the core, or dial "Bright Saturation" right down + setting saturation to 100% or lower, in order to keep the Color module from introducing/recovering color in the highlights. You can also use a starmask and the Layer module to do this selectively for just the Star cores.

As I said, it's a non-trivial amount of post-processing, ideally avoided by addressing the issue during acquisition, but it can be done/mitigated in post-processing!

Does this help?
Ivo Jager
StarTools creator and astronomy enthusiast
alacant
Posts: 211
Joined: Mon Jan 25, 2016 7:03 am

Re: fringe killer filter add blue back to central star

Post by alacant »

Thanks Ivo. These were the bits I was missing:
--- Deconvolution

Parameter [Radius] set to [3.2 pixels]
Parameter [Iterations] set to [1]

Now you can use StarTools' unique "Restore" functionality to restore the data to a state where it is unstretched (so we can merge it with the unstretched red and green channels), but with deconvolution still intact (the Restore functionality is a perk of the Tracking feature of StarTools, allowing you to "time travel" and pick and choose aspects of the image you like - an "undo" feature on steroids). We choose "linear, wiped, deconvolved".
This is with bin 35% and decon radius at 2.8. It works well. I don't think I'll need to point out which is which but as always with apologies for my total lack of skill with the st colour module.
Thanks again and clear skies,
Steve

** Ivo, do we have anything like Linux bash or sh for StarTools? It would be great to be able to script stuff like this.
ngc6888.jpg
ngc6888.jpg (224.1 KiB) Viewed 16385 times
6888.jpg
6888.jpg (197.13 KiB) Viewed 16385 times
User avatar
admin
Site Admin
Posts: 3364
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: fringe killer filter add blue back to central star

Post by admin »

That's an awesome demonstration of the effectiveness of the technique Steve!:thumbsup:
** Ivo, do we have anything like Linux bash or sh for StarTools? It would be great to be able to script stuff like this.
The thing with stuff like this, is that different images require slightly different parameters/treatment. Much depends on the night (seeing), focus, collimation, amount of subs, even the way the data was stacked.
StarTools is all about automating things that can be safely and objectively automated by simply being more clever about how your data, actions and input are (re)used and analysed.
For matters of aesthetics and things that are subjective, old fashioned user input is still required and desirable (un?)fortunately.

Hope that makes sense!
Ivo Jager
StarTools creator and astronomy enthusiast
User avatar
admin
Site Admin
Posts: 3364
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: fringe killer filter add blue back to central star

Post by admin »

Actually... If you have any more data you can throw my way - especially with stars on/in nebulosity I may have some other ideas on how to tackle this in a more automated way...
Ivo Jager
StarTools creator and astronomy enthusiast
alacant
Posts: 211
Joined: Mon Jan 25, 2016 7:03 am

Re: fringe killer filter add blue back to central star

Post by alacant »

Hi Ivo. Yes, you make a lot of sense. I must also admit that the method has taken many hours of frustration over several weeks to get to this stage and all credit to you for making it possible. Your point about so many variables is very valid: the decon radius is critical. Dunno. I'd like to think the crescent nebula is an example of worst case scenario with all those bright stars so as to be able to set a sensible upper limit on the blue channel decon.... M42 is coming around again and I remember having trouble with the bright stars there too. I too doubt whether there's a one size fits all possibility.

I think we've also got to take into account fashion. It's currently hard to find a nebula taken without the stars obliterated by a H alpha filter or with everyone trying to recreate the Hubble pallate and in the process spending more on one filter than my entire setup!

But back to reality and affordable refractors on rusty mounts. Tonight it's the pelican nebula or bust. Blue halos or not. I'll try to recreate as much of what I did as possible and report on that.

Thanks again for your patience,
Clear skies
User avatar
admin
Site Admin
Posts: 3364
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: fringe killer filter add blue back to central star

Post by admin »

alacant wrote:Hi Ivo. Yes, you make a lot of sense. I must also admit that the method has taken many hours of frustration over several weeks to get to this stage and all credit to you for making it possible. Your point about so many variables is very valid: the decon radius is critical. Dunno. I'd like to think the crescent nebula is an example of worst case scenario with all those bright stars so as to be able to set a sensible upper limit on the blue channel decon.... M42 is coming around again and I remember having trouble with the bright stars there too. I too doubt whether there's a one size fits all possibility.
I worked on another method this weekend that may yield better results again, recovers/retains star color and has the potential to be automated. I just need some more data to test with... If you would have any, it would be greatly appreciated!
I think we've also got to take into account fashion. It's currently hard to find a nebula taken without the stars obliterated by a H alpha filter or with everyone trying to recreate the Hubble pallate and in the process spending more on one filter than my entire setup!

But back to reality and affordable refractors on rusty mounts.
This is exactly why I don't want to let go of this one yet (and the reason why StarTools exists). Achromats give a lot of bang for your buck.
Tonight it's the pelican nebula or bust. Blue halos or not. I'll try to recreate as much of what I did as possible and report on that.

Thanks again for your patience,
Clear skies
Looking forward to your results and findings!
Ivo Jager
StarTools creator and astronomy enthusiast
alacant
Posts: 211
Joined: Mon Jan 25, 2016 7:03 am

Re: fringe killer filter add blue back to central star

Post by alacant »

Hi Ivo
Thanks for taking an interest in the lower end equipment, all too often rejected for astrophotos.

No luck with the pelican last night; new local bar's opening night with free beer won through I'm afraid;)
I just need some more data to test with... If you would have any, it would be greatly appreciated!
How about this? https://drive.google.com/open?id=0B_uUt ... EgtbHB3R0E
HTH,
Steve
alacant
Posts: 211
Joined: Mon Jan 25, 2016 7:03 am

Re: fringe killer filter add blue back to central star

Post by alacant »

Update
Getting better. Chanced upon an even cleaner route. I placed a uv-ir filter** instead of a lp filter*. It helps reduce the halos -a lot-, needing less decon and no reduction in the blue luminance. If we could get the fringe killer to heal against the nebula, wow. But hey, I've seen expensive APOs struggle with that.

**Same effect with both filters in the light path.

* Why does this work? I'm wondering if too many/wrong filters have been removed from the camera. The blue histogram is moved to a lower intensity as if the uv or ir -or whatever frequency is being cut- was being recorded as visible. Is that possible?

TIA and clear skies,
Steve
Attached a snap of the filter and a before and after screenshot.
uvir.jpg
uvir.jpg (7.54 KiB) Viewed 16322 times
tulip.JPG
tulip.JPG (160.97 KiB) Viewed 16322 times
Post Reply