HaLRGB

[jeffweiss9]

Old topic but, since Juan's Narrow Band / Wide Band combination module isn't ready, I was going to try Jack Harvey's HaRGB prescription (2008 Jan 24).  My question is: does it make sense to do this with a real (processed) luminance data channel rather than the luminance channel extracted from a processed RGB image? (I've already got the Ha L R G B data channels).   Same question for Juan's method with the additional question of doesn't mixing real luminosity data with the Ha luminosity data in some way help the image over just using Ha alone as the luminosity?
 
Thanks,
-Jeff

[lucchett]

Hi,
Assuming your L data is deep and your color data is binned, i think you can process your image as LRGB, then split the rgb channels of this image and apply Jack's method or other blending way you want to try.
Andrea

[jeffweiss9]

Thanks, Andrea-
   I did try to mix the Ha with the Red in the LRGB but the problem was mixing Ha and L in the luminosity channel.  In my case, the L was pretty deep (I was at a very dark site) but the Ha just showed up a lot more H2 cloud than was visible in the L channel. (In fact, I had to search a bunch of Ha Bubble images to even find it in other posted NGC 7635 images).    I did try to simply add Ha+L (1 hour each) for a luminosity channel in PI, but not much of the "extra" H2 cloud came through.  I note Jack Harvey's prescription (I believe) is to make the L(Ha-R)GB and then just add the Ha to that.  Is that what is recommended?
   To be more explicit, here's the image I was after after cropping (needs lots of rework):

http://www.astrophotogallery.org/member-galleries/p6136-ngc7635-bubble-nebula.html

Here's the Ha image uncropped that shows all the extra Ha cloud that wasn't visible in any other channel:

http://www.flickr.com/photos/9906726@N08/4966288963/#/photos/9906726@N08/4966288963/lightbox/

So the questions was how to keep that stuff in the final image?

Thanks,
-Jeff

[Jack Harvey]

Well my method with the Lum substitution is a couple of years old.  Here is Vicent's fairly recent suggestion which I now use with good effect.

in PixelMath

Red Master - Ha Master = img 1

Ha Master - Img 1*(7/100) = Ha Master (Vicents method)

Now Combine above Ha Master with Red Master in some Proportions with Pixel Math (ie 60% Ha and 40% R) to make a HaR Master then use the LRGB tool put the newly combined HaR in red and G in green and B in Blue.  I don't shoot L so that is up to you and would go in L but maybe not at 100%???

use 7 as the band pass for Ha is 7nm and 100 as the band pass for Red is 100 -CAHA Baader filters

AstroDons use 5 for Ha and 100 for Red

Buena Suerte, Hope you followed that<G>

[jeffweiss9]

Thanks for the update, Jack-

If I've understood, this boils down to:

 HaMaster'  = 1.07*HaMaster -0.07*RedMaster       (I have Astrodon LRGB but Orion narrowband 7nm Ha)

 HaRMaster = 0.60*HaMaster' + 0.40*RedMaster        (or other admixture. )

These operations then reduce to

 HaRMaster = 0.64*HaMaster +0.37*RedMaster)

 which then is used in LRGB (re)combination in place of RedMaster.

 If I have shot L (my RGB was 2x2 binned and relatively short total integration time), then I should just mix it in with the L from the RGB at some percentage less than 100%.

 That's pretty much what I did, using 0.30*HaMaster + 0.70*RedMaster for the HaRMaster (thought it looked better than 0.50/0.50 or 0.70/0.30) and 0.50*HaMaster+0.50*Lum for the Luminosity channel in the LRGB combination (ignoring the L that came from the RGB).  

 I did discover a major goof in applying DBE to my Halpha data.  Somehow it generated those "H2 clouds" which, in fact, appear to be spurious as trying DBE again on the original data, they are gone:

http://www.flickr.com/photos/9906726@N08/4989303952/#/photos/9906726@N08/4989303952/lightbox/

That still looks much better/different than either the L or R data, but I'll have to start again with this corrected Ha with something like this procedure.    Since I have very different exposure times for my Ha and R, however, do they need to be equalized in some way (linear fit?) before mixing?

  Thanks,
Jeff

[Jack Harvey]

"HaMaster'  = 1.07*HaMaster -0.07*RedMaster       (I have Astrodon LRGB but Orion narrowband 7nm Ha)"

You are close.  1.07 should be 0.07.  The 1 is actually from what Vicent calls Image 1

[jeffweiss9]

Jack-
 You lost me there, as it clearly must not be algebra.
Can you direct me a reference?  I've succeeded in getting completely confused by trying to piece together the various threads on HaRGB or HaLRGB. If the Ha and R, for example, have different exposures (and maybe binnings), doesn't some sort of equalization of intensities have to be done before these Pixel Math operations?
Thanks,
-Jeff

[Nocturnal]

Jeff,

PM will upscale images to fit, as far as I remember. Give it a try and if it doesn't work, upscale the images yourself before applying PM. Yes, I would assume that both Ha and R need to have the same linearity or sub exposure length. Total exposure affects SNR and not linearity so that doesn't matter for PM. You could use the new linear fit process.

Disclaimer: I never do this myself, proceed at your own risk

[Jack Harvey]

As I stated, this is a method developed by Vicent and he passed it on to me.  IF you look at my picture, do you see a math major?  Nope, I am a cowboy<G>.  For further enlightenment I suggest a)  Vicent Peris  b)  Juan  c)  Stephen Hawking  d)  all of the above  And of course the correct answer is d.<G>.

Seriously Jeff I have used this several times on images and it works well for me.  Good Luck and hopefully Vicent can chime in.

[Jack Harvey]

FWIW here is one I finished just now with that technique applied to both the Ha and Oiii with a final combination of HaHaR OiiiG OiiiBOiii.
[Yeah I know it is uncropped and a couple of stars need some attention-it is not a final ver. yet<G>]

Helix Nebula

http://www.starshadows.com/_img/image/gallery/2/HelixNebula_final.jpg

[sleshin]

Jack, what an incredible image. Look forward to the final version.

Steve

[jeffweiss9]

That is an amazing(!) image, Jack-

Learning more about HaLRGB but still pretty confused by the various techniques... While waiting for a new PI tool ,  I fed all the PixInsight-processed components to Tony Hallas' HaLRGB technique where he assembles everything in, dread, PS:

http://www.astrophotogallery.org/member-galleries/p6136-ngc7635-bubble-nebula.html

It may not be the best result but it is better than I was able to do with simple Ha-R and Ha-R combinations in Pixel Math.  Also it did succeed in meeting Hallas' HaLRGB goals of getting everything available out of the Ha Luminosity as well as from the "Ha as Red" to provide the color support for the luminosity. Now if someone could only automate that in a PixInsight script, I'd be happy to give them the his prescriptions, including the equations I found for the blending modes involved so they could be translated into Pixel Math.  

Clear skies,
-Jeff

P.S. Here's the full image with M52 (I was going to post the Ha, but Hallas' HaLRGB technique got it all):

http://www.astrophotogallery.org/member-galleries/p6239-ngc7635-bubble-nebula-and-m52.html

[Jack Harvey]

Nice work.  Now that you have good color I bet you can pull out some better resolution.  Try either the HDR tool with luminosity checked and the rest at defaults.  OR use Atrous wavelets.  Enable the live preview so you can see the result.  Use defaults and select first layer.  Use bias of 0.1-0.2 and check nosie reduciton at .5 and deranging.  Layer two bias of 0.05 to .1 is my guess and then noise at .3 and deranging.  Maybbe level 3 at same as level two.  Watch the preview and move the sliders for best result.  After applying zoom in and check that you did not induce too much noise and if so use ACDNR.

[jeffweiss9]

Thanks, Jack-
 Tried ATWT with settings like that and it definitely sharpened things up a bit.
Images updated at the same links.
-Jeff

[lucchett]

Hello Jeff,
can you tell us a bit more about the Hallas method?
It would be interesting to compare with others technique shared in the Forum.

By the way,
I think it's time for the Pixinsight Guru's to produce in a jointed effort a new tutorial in Ha blending.!:-)


Andrea