Correcting a strong...
 
Share:

The problem with the data upload limit for attachments has been fixed. I have restored it to 30MegaBytes. A recent forum software upgrade was responsible for the changed limit. Please accept my apologies for missing this when upgrading the forum sofware, Mabula.

Correcting a strong Blue/Violet Cast  

  RSS

(@mestutters)
Red Giant Customer
Joined: 2 years ago
Posts: 58
24/09/2018 12:54 pm  

Hi Mabula/APP users,

I wonder if anyone can offer some advice on how best to correct a strong blue/magenta cast that I am getting while processing an HaLRGB image of Crescent Nebula.

Attached are screenshots of my APP work in progress a)  after Light Pollution Correction and b) Star Color correction.

a) 

Screenshot 2018 09 24 09.20.08

b)

Screenshot 2018 09 24 10.08.12

After LPC I think the image is looking promising although it clearly needs some further work to improve the color.

During CSC however the image has taken on a strong Blue/Volet cast although the star colors look about right to my eyes and also according to tbe color balance charts as I understand them.

My problem is that with APP's HSL selective color module I am unable to correct the background colors to get anywhere near an acceptable result.

My questions therefore are:

a)  Is it possible to use HSL module to reasonably correct the color cast and if so how should the problem be approached?

b)  If HSL is not able to make a sufficient correction what should I be looking to achieve at the (?) RGB combine stage in order to get a better starting point for subsequent adjustment?

I am reprocessing this image after acquiring some extra data so I know with effort a resonable is result is achievable although even with the below image I had to resorted to Photoshop for final substantial color balancing.

combine RGB image v04.2.05 lpc cbg csc SC St

I am happy to provide my FITS files if anyone wishes to have a try.

Regards

Mike

This topic was modified 11 months ago by mestutters

ReplyQuote
(@vincent-mod)
Galaxy Moderator
Joined: 2 years ago
Posts: 551
06/10/2018 7:56 am  

That might be a good idea (to share the files), the star colour graph looks really weird and there seems to be some over representation of green in the data.


ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 2 years ago
Posts: 2106
14/10/2018 4:27 pm  

Hi Mike @mestutters & Vincent @vincent-mod ,

Yes, I see the issue with the data. My first guess would be that the HaLRGB composite actually is the problem. If I look at the histogram in the Remove Ligh Pollution stage. Green versus Red+Blue is rather unbalanced and this most likely leading to the issue in star color calibration.

The star color calibration Black Body models need broadband data to work well, but you have supplied broadband with R,G,B data. So I suspect the issue is to be found in how the Ha data is actually added to the composite.

Regarding HSL selective color: it's needed for final small tweaks mostly. We first should get a better result at the star color calibration stage I feel... before continuing with processing.

I would be happy to have a look at the data, please send the integrations per channel to me. I will then have a look to find a possible issue and otherwise suggest a better workflow 😉

You can send the data to support@astropixelprocessor.com using wetransfer.com or dropbox 😉

And I agree, the data itself looks promising, the nebulosity looks rich in signal.

Kind regards,

Mabula

Main developer of Astro Pixel Processor and owner of Aries Productions


ReplyQuote
(@mabula-admin)
Quasar Admin
Joined: 2 years ago
Posts: 2106
15/10/2018 9:09 pm  

Hi Mike @mestutters ,

Thank you for sending me the data.

When I load the R,G,B channels into the RGB Combine tool, I immediately get an odd result I think.

RGB composite

This is just with R,G,B multipliers at 1.

The magenta cast comes from Red and Blue dominating the green channel. The nebulosity is magenta everywhere which is odd.

If we look at the Red, Green and Blue channels:

Red
Green
Blue

We can see why, the nebulosity and Cresent is visible in Red AND Blue, not in Green.

I would almost think that somehow red and blue data was mixed in the final red and blue integrations.

The brightest stars in the red channel are also the brightest stars in the blue channel it seems and there is a big difference in quality as well.

Perhaps to investigate further, can you share some individual red, green and blue frames ? Maybe that will give a better understanding of what is happening here 😉

Kind regards,

Mabula

 

 

 

 

Main developer of Astro Pixel Processor and owner of Aries Productions


(@mestutters)
Red Giant Customer
Joined: 2 years ago
Posts: 58
30/10/2018 9:40 am  

Hi Mabula,
Thank you for taking a look at my earlier attempts at Crescent nebula image and for your suggestion as to what might be wrong. I took a close look at my all my subs and found that amongst my Blue frames I had several (c30% of total) that looked as if they had actually been recorded through Green filter rather than Blue. Not sure how that happened but an operator error I guess!
Anyway I managed to capture a few extra Blue subs the other evening to replace the faulty ones and have had another shot at processing. The result now looks rather better but on the B-G vs G-R chart the star profiles seem to have a negative slope. I think that this cannot be right although on other images I have done I generally have to set the slope value to near the lowest settings available.
I hope I have not made another silly mistake but I would much appreciate your view as to where I might still be going wrong.
I will very shortly forward my latest integration frames and examples of my HaLRGB frames.
Best Regards
Mike
PS : I've just enjoyed viewing your AstroImaging Channel talk and am looking forward to seeing the further improvements you have planned for APP. Hope you very much enjoy your holiday too.


(@mestutters)
Red Giant Customer
Joined: 2 years ago
Posts: 58
04/11/2018 9:35 am  

Hi Mabula/Vincent,

Please consider this problem closed.  I managed to capture a few more subs on this target and was then able to carefully compare what I had with the new results.   After weeding out a few more older subs that looked suspisiously as if they too were incorrectly labelled  I am now getting a result in CSC that looks about right. 

Thanks for your time and patience

Mike

Screenshot 2018 11 04 08.18.02

 

 


(@vincent-mod)
Galaxy Moderator
Joined: 2 years ago
Posts: 551
04/11/2018 9:36 am  

Was just about trying to be of assistance. 🙂 Glad it worked out ok!


ReplyQuote
Share: