Combining Ha data with LRGB?

[twade]

To all,

I'm still trying to work out a legitimate work flow for combining Ha data with LRGB data without the "pink" effect.  I attempted Vicent's technique but it doesn't seem to scale very well when working with wide fields of view.  At this time, my hope is to create two substitute channels using the Ha data in the Red and Blue channels. Basically, there's always a percentage of H-beta present in Ha regions. The H-beta generally ranges from 10-30 percent of Ha depending on attenuation due to intervening dust and gas.  The difficult part is guessing the proper percentage of H-beta to use for a particular image.  Here's my work flow so far:

1.  Since there's less signal in the Ha image, I try to match the brightness of a non-saturated star in both images.  This is done using a mid tone transfer function (HistogramTransformation).  I assume this destroys the linearity of the data, but I don't see any other way to get their intensities to match reasonably well.

2.  Once their star intensities match, I use PixelMath to create a "blue" channel using the synthetic H-beta derived from the H-alpha image .  For example,

Blue*0.74 + Ha*0.26 --> New Blue channel with synthetic H-beta (26 percent)

3.  I then do a 50 percent combine of the Red and H-alpha images:

Red*0.50 + Ha*0.50 --> New Red channel using H-alpha data

4.  I combine these two new channels with the original Green channel to produce a RGB image.

5.  I then neutralize the background using:

$T - Med(bkg) + 0.025

6.  I then use a G2V star within the field of view to correct any color casts.  I realize G2V calibration is not the most ideal solution; nevertheless, this is all I have to work with at this time.   Any better suggestions would be greatly appreciated.  For example, using PixelMath:

RGB/K = 1.3216
G=1.3387
B=1

In my initial attempt, I have a pinkish result due to choosing a H-beta percentage that was too high.  I'm next going to try a value around 18 percent.  Another artifact I see is I've lost my star colors.  Is this common when combining Ha data into a RGB image?  How can it be avoided?

Any comments and/or help would be greatly appreciated.

Thanks,

Wade

[Jack Harvey]

Chris Schurr has used a technique combining Ha wiht both Red and Blue channels for some time and many have had good sucess.  He reported the details of his technique at AIC I think in 2007.  HIs AiC presentations are fond on his web page.  http://www.schursastrophotography.com/

I personally have not used it but know guys tht do and get great results.  He uses PS but with some workaround I am sure PixI can also do it.  HTH

Jack

[twade]

Jack,

> He reported the details of his technique at AIC I think in 2007

I downloaded the presentation and looked over it.  It looks pretty easy to get it to work with PI.  It appears, he completely stretches his channels before attempting to combine the Ha data.  Although, he doesn't go in to any details on subtracting the stars from the final combination, hopefully I can find a way to do this in PI.   Do you have any ideas?

Wade

[Jack Harvey]

I do not at present but bet Juan or others could.  I suspect a starmask and then some sort of Pixel Math might do it?  Chriss is a very helpful guy so I would not hesitate to email him for some help.

[Jack Harvey]

Finally had some more time to mess with this dilemma.  Here is what is working pretty well for me.
1.  Process your RGBs into a finished RGB.fit
2.  Process your registered Ha into a stretched and finished Ha.fit
3.  Use Extract Channels and extract the R,G,B channels
4.  Use Pixel Math to combine R and Ha   ie     Eta rgb_R+(0.35*Eta ha)
    Above is for a 35% Ha and you may want to try more
5.  Repeat with blue frame using about 15% ha mix
6.  Using the LRGB combine
    HaR in Red
   Green in Green
   HaB in Blue
7.  Combine and finsh processing.

Let me know how this works for you.  Obviously a variation of Shurs tech

[Harry page]

Hi Gents

Have had a go as suggested with fairley good results ( still a bit pink)
Would you apply the ha to a RGB image and save the lum to last or apply the HA to LRGB


Regards Harry

[Jack Harvey]

Some do the latter and overlay the Ha on the LRGB but my best results have been using RGB and Ha.  Here is Eta done that way.

http://www.starshadows.com/gallery/display.cfm?imgID=276

Jack

[vicent_peris]

Hello all,

the is a key point in this processing area: making a profitable use of our narrowband data. This must be understood in two different ways:

- Recovering the signal to noise ratio of nebular structures. This is of special importance for urban imagers, where broadband data is very noisy!

- Enhancing the emission line structures over the stars.

With the method you talk about, you are using only (as Jack says) a 35% of the H-alpha in the red channel, and only a 15% of it in the blue channel. Furthermore, this combination has a little impact on the luminance of the image.

The method I propose in my article makes, at first, a substitution of the red channel with *exclusively* h-alpha data. The second part of the algorithm introduces a regulated amount of h-alpha data into the luminance, with near zero color shift.

I think you should keep in mind these two points when combining these kind of images.

I also think we must do a new module fo HaRGB combination.


Regards,
Vicent.

[Harry page]

Hi

Yes I think a Ha RGB modual would be very desirable, and I think someone somewhere! promised one quite a while ago.


Regards Harry

[Jack Harvey]

Vincent and Harry  I agree I used a very conservative addition of Ha to my red and blue channels.  My experiience is different images accept different amounts of Ha to give pleasing results.  I hope if we do get a tool to combine Ha with the wide band data that it includes sliders that allow varying amounts (0-100%) to each channel.  If you apply 100% to the lum channel then you have an Ha overlay on the RGB which some people like to do, or you can apply varying amounts to the Red, Green and Blue channels with or without the overlay.

[Harry page]

Hi Jack

You must have red my mind on the sliders! also there would need to be a option to either add or replace the desired amount on each channel.
Used with the preview option it would be very powerful and a good addition to the pixinsight world


Regards Harry :lol:

[twade]

Vicent, Jack, and Harry,

I appreciate all your input and help.

I've tried all the suggestions, but I still can't get a satisfactory result.  Perhaps, I'm not exposing the R, G, B data enough.  Would any of you be interested in seeing what you can do with my data?  I'll post it to the PTeam directory.  I think the more wide-field you go the tougher it is to get things to work properly; however, I'm sure its my lack of processing skills too.

I'll go ahead and post it to the twade directory.  It will be compressed using WinZip.  If you have any problems, please let me know.

Wade

[twade]

To all,

The file, M45_105mm.zip has been uploaded to the PTeam directory under twade.  If interested in trying to combine the individual L, Ha, R, G, and B channels into an image, you will need to download all eight files (i.e. the main M45_105mm.zip and its seven parts).  Total size is 196 MB.

Have fun!

Thanks,

Wade

[Jack Harvey]

Wade I will have to wait a coule of weeks as this starts the "dark period" and that is when we get busy at CTIO for  a couple of weeks.  Hopefully someone else can jump in and get you an answer sooner.

[Juan Conejero]

Hi Wade,

I'll download the file and will play with the images tomorrow. Thanks for uploading your data!