Page 1 of 1

Creating a synthetic green channel

Posted: Thu Jul 11, 2013 9:09 pm
by cybereye
Ivo,

After our little dscussion regarding my missing green channel because of my UHC-S filter, I started reading up on creating a synthetic green from Ha and OIII data. As this is essentially what I'm dealing with, is it possible to have StarTools somehow automate the creation of this green channel?

Cheers,
Mario

Re: Creating a synthetic green channel

Posted: Sat Jul 13, 2013 1:13 am
by admin
cybereye wrote:Ivo,

After our little dscussion regarding my missing green channel because of my UHC-S filter, I started reading up on creating a synthetic green from Ha and OIII data. As this is essentially what I'm dealing with, is it possible to have StarTools somehow automate the creation of this green channel?

Cheers,
Mario
Well, with 'Channel Interpolation' on, the LRGB module creates a synethetic red, green or blue channel based on which channel is missing.
Alternatively, you could mix your own green channel from different sources using the Layer module (in grayscale), save the (grayscale) result and subsequently use the LRGB module to load that result as the green channel.

Does this help?

Re: Creating a synthetic green channel

Posted: Sun Jul 14, 2013 8:56 pm
by cybereye
I separated my image into it's 3 components but found the green channel generation in the LRGB module to be a tad too strong. How does the green channel get created? With my playing around, and with some searching on the web, the best method I've found for creating the green channel is to layer the blue over the red channel in GIMP/PS and set the blue layer to "multiply" rather than "normal". The resultant doesn't seem to be as strong as either the blue or the red channels but gives a fairly pleasant result when all 3 are recombined.

As I have an OSC I need to separate my channels in PS (GIMP doesn't handle 16bit tif files although I have a 2.9 version to try...). This is not easy to do. Does StarTools have the capability for you to separate the 3 channels, ditch the green and then synthesis it using the multiply method I mentioned previously in this post?

This option, I believe, would be a great help to we "Urban Astrophotographers" who use LP filters... well it would to me at least!!

Cheers,
Mario

Re: Creating a synthetic green channel

Posted: Sun Jul 14, 2013 11:16 pm
by admin
cybereye wrote:I separated my image into it's 3 components but found the green channel generation in the LRGB module to be a tad too strong. How does the green channel get created?
The green channel is created by averaging the blue and red channels, which is the same as layering red on top of blue with an opacity (or 'Blend Amount' in ST) of 50%.
With my playing around, and with some searching on the web, the best method I've found for creating the green channel is to layer the blue over the red channel in GIMP/PS and set the blue layer to "multiply" rather than "normal". The resultant doesn't seem to be as strong as either the blue or the red channels but gives a fairly pleasant result when all 3 are recombined.

As I have an OSC I need to separate my channels in PS (GIMP doesn't handle 16bit tif files although I have a 2.9 version to try...). This is not easy to do. Does StarTools have the capability for you to separate the 3 channels, ditch the green and then synthesis it using the multiply method I mentioned previously in this post?
I never heard of this method before, but ST can certainly do this. The LRGB module, when confronted with a colour image, extracts the relevant channel (e.g. if you click 'Red' and load a colour image, the LRGB module will only extract the red channel from that colour file). Using this, you can extract the relevant channels from a colour file.

So, to create a multiplication of the two channels;
Launch LRGB, click red, load image, red channel will be extracted, Keep (we're not interested in Tracking right now, so I just indicated the image has been stretched to get Tracking off my back :) ).
Launch LRGB, click blue, load image, blue channel will be extracted, Keep (again, we're not interested in Tracking right now).
Launch Layer module, 'Fill mask', click 'Undo > Bg' or 'Undo > Fg' to copy the undo buffer (which will contain the red data) into the foreground or background (doesn't matter for multiply operation). Set 'Layer Mode' to 'Multiply'. That's your new green channel. Click 'Keep', Save your new green channel.

Now to you simply use the LRGB module again to load R and B from the original image and load the new green channel we just made as G.
This option, I believe, would be a great help to we "Urban Astrophotographers" who use LP filters... well it would to me at least!!
I'd love to see your results as, to be quite honest, I doubt it would yield anything close to a correct colour. If you do go this route, however, I would use a synthetic luminance frame as well, so that your exposure time in the green channel doesn't go to waste. In this case, you can simply click the 'L' button and load the original image as the luminance file. The LRGB module will automatically average all the channels and create a luminance frame out of them.

There is one other thing you could do though to attain correct star colours; simply image without an LP filter for your RGB (colour) data, and use a blend (depending on exposure time) of the LP filtered data + non-LP filtered data as your luminance data. It may seem a bit tricky at first, but once you got the hang of it it should yield the best of both worlds in your situation.

Re: Creating a synthetic green channel

Posted: Mon Jul 15, 2013 2:35 am
by cybereye
Ivo,

Here' what seems to be a powerpoint presentation from the 2007 AIC for the technique I used. It may be of some interest...

http://aicccd.com/archive/aic2007/Cannistra.pdf

Cheers,
Mario

Re: Creating a synthetic green channel

Posted: Mon Jul 15, 2013 3:05 am
by admin
cybereye wrote:Ivo,

Here' what seems to be a powerpoint presentation from the 2007 AIC for the technique I used. It may be of some interest...

http://aicccd.com/archive/aic2007/Cannistra.pdf

Cheers,
Mario
Apologies Mario. I should learn to *read*. :doh: You are looking for a method to create a synthetic green channel from narrowband data (not RGB data). In that case any talk about the 'correct' color balance is moot and the description in this document may well work to your satisfaction. What does stil stand though is my suggestion for an RGB solution in the face of the light pollution you're experiencing!

Re: Creating a synthetic green channel

Posted: Mon Jul 15, 2013 4:05 am
by cybereye
Ivo,

Just to clarify what I am after... I am trying to recreate the green channel from my OSC. As you pointed out with my Trifid image, data from my OSC CCD is essentially missing the green channel. In essence, the UHC-S filter only let wavelengths through around the Ha and OIII areas giving me a bicolour image. To compensate for this I need to separate the data into the blue and red channels - giving me my OIII and Ha data - and then "restore" the green channel from this. From what I can gather, I won't get yellow stars back, but I'm hoping that it will get rid of the red halos.

Does this make sense?

Cheers,
Mario

Re: Creating a synthetic green channel

Posted: Mon Jul 15, 2013 4:31 am
by admin
cybereye wrote:Ivo,

Just to clarify what I am after... I am trying to recreate the green channel from my OSC. As you pointed out with my Trifid image, data from my OSC CCD is essentially missing the green channel. In essence, the UHC-S filter only let wavelengths through around the Ha and OIII areas giving me a bicolour image. To compensate for this I need to separate the data into the blue and red channels - giving me my OIII and Ha data - and then "restore" the green channel from this. From what I can gather, I won't get yellow stars back, but I'm hoping that it will get rid of the red halos.

Does this make sense?

Cheers,
Mario
I think the guide for creating the green channel was written for narrowband data only. Your filter is a wide band filter allowing through pretty much everything (Ha, Hb, S2, O3, N2, etc.) *except* the mercury vapour and sodium vapour wavelengths.
Narrowband renditions are false color by default. For instance OIII should really have a greenish appearance, rather than pure blue (as mapped in the tutorial/article you linked to), so killing your green channel is also going to be killing much of your OIII data that's mixed in there. S2 and Ha are both red for instance, that's why you would want to map them to distinctly different channels to emphasise their detail.
You are trying to apply a narrowband technique to RGB imaging, which, I hate to say it, is not going to work!

Your best bet is to do separate imaging of some RGB data without your lp filter, to get the colors right (they can be recovered). Then do the bulk of your exposure with your filter in place and create a synthetic luminance frame out of both exposures. Combine synthethic luminance frame with your RGB data and you should have correct colours.
You need less exposure to get colours right than for luminance data which does require longer exposures. The human eye is less sensitive to defects/noise in colour than it is to defects/noise in luminance data. It's the whole premise behind most lossy video and image codecs, and is also the reason why astrophotographers can get away with using bin 2x2 for RGB, but use bin 1x1 for their luminance data. :)

Re: Creating a synthetic green channel

Posted: Mon Jul 15, 2013 5:00 am
by cybereye
Ivo,

I see what your saying. When I look at various filter responses I see my problem as a "Broad Narrow Band" imaging problem where I have a greatly reduced green channel. I'll still go through the motions as you've described before for completeness as I still think this method may just help... I'll let you know how I go! Otherwise I'll give your no filter imaging technique a go, although it's not my prefered option... :)

Thanks once again for all your help,

Cheers,
Mario

Re: Creating a synthetic green channel

Posted: Mon Jul 15, 2013 5:16 am
by admin
cybereye wrote:Ivo,

I see what your saying. When I look at various filter responses I see my problem as a "Broad Narrow Band" imaging problem where I have a greatly reduced green channel. I'll still go through the motions as you've described before for completeness as I still think this method may just help... I'll let you know how I go! Otherwise I'll give your no filter imaging technique a go, although it's not my prefered option... :)

Thanks once again for all your help,

Cheers,
Mario
No problem. Let me know your findings! We may yet learn something from this...