Losing Hair Over Deconvolution

[SeattleStarGazer]

I'm working on a bicolor narrowband image of the Horsehead, blending H:O:O in a straight 1:1:1 ratio. When I use decon on the monochrome H-a image, I get pretty good results with a local deringing star mask and lum mask adjusting the global dark and iterations.However, as soon as I convert to RGB color and re-run with the same basic settings, things go to heck. Huge dark rings and lots of white noise. I don't do any processing other than blending. Am I missing a key concept somewhere?

Thanks!

Wayne

[eganz]

Since you been successful using deconvolution on the hydrogen image, The 1st thing to try would be to use deconvolution on your oxygen image by itself.

[jkmorse]

Decon is one of those processes where each image requires you to tweak the settings to get it right.  When you introduce the added data from the combined image, things change and it requires you to change as well.  But Egnaz is on the right track in that I always find it easier to work on a monochrome image with Decon than a combined image.  That doesn't mean you are out of luck though.  There are a couple of ways to tackle this:

1.  Use the combined image and tweak your settings.  You most likely have to fix the local support slider to get rid of the rings, going really low (say 0.002) and slowly working your way up;

2.  Take the combined image, use Channel Extraction with the Lab setting.  This will give you three images, the Luminance frame and the a and b color channels.  Put the color images aside (but don't close them as you will need them in a bit).  Now take the L image and make a clone (simply left click on the name in the upper left corner and pull a new image over to the right.  Its best to experiment on a clone, not the original).  Now you have extracted the L image which is where all the detail is and you can run Decon on it.  Hopefully working on the monochrome will make the Decon process easier.  Then, when you get what you want, simply open Channel Combination, select Lab and recombine the three images.

Hope that helps.  By the way, I do have a 70+ page workbook that goes into a lot of this, including details on Decon and the other major processes, workflows, etc, etc.  I share it freely with around 200 people around the world.  If you are interested, just drop me a line at jkmorse57@gmail.com.

Best,

Jim 

[SeattleStarGazer]

Thanks guys! I gave a try to deconvolving the mono images separately. In fact, the O-III image took very different settings to be successful, so I suspect when combining them first the differences would significantly impact the process. The O-III image is dimmer and noiser than Ha, so I think it would be a good idea to get more O-III data too. But I'll give that lab color process a go too.

[Geoff]

Decon is one of those processes where each image requires you to tweak the settings to get it right.  When you introduce the added data from the combined image, things change and it requires you to change as well.  But Egnaz is on the right track in that I always find it easier to work on a monochrome image with Decon than a combined image.  That doesn't mean you are out of luck though.  There are a couple of ways to tackle this:

1.

2.  Take the combined image, use Channel Extraction with the Lab setting.  This will give you three images, the Luminance frame and the a and b color channels.  Put the color images aside (but don't close them as you will need them in a bit).  Now take the L image and make a clone (simply left click on the name in the upper left corner and pull a new image over to the right.  Its best to experiment on a clone, not the original).  Now you have extracted the L image which is where all the detail is and you can run Decon on it.  Hopefully working on the monochrome will make the Decon process easier.  Then, when you get what you want, simply open Channel Combination, select Lab and recombine the three images.


Best,

Jim

I would have thought that ensuring that the target in the deconvolution panel is selected as "Luminance (CIE Y)" would have done the same thing as splitting the channels and then recombining after decon.

[jkmorse]

Geoff,

You may very well be right, but I have never done it that way.  I have always only worked on a lum image.

Jim