Color Channel Alignment

[niteman1946]

Hi Group,

Here is my narrowband version of a section of NGC2237.  The histogram on the Astrobin window shows that the red, green and blue are in poor agreement.  I saw the same thing during processing with PI.

Anyone have comments or suggestions on the cause and potential fix?

Thanks,

Mark

[sreilly]

I almost always see this with narrow band images. What I do is align using each color, instead of the RGB setting in PI with the Histogram tool, choose jut the color that is off and align using the sliders. It will take a minute or two to see what direction you need to go without clipping the data.

[sreilly]

And the last,

[Philip de Louraille]

Well narrow band is narrow band. The data you obtain is physical in nature, not thermal. (Ha, for example, is due to electrons going down from n=3 to 2.)
So there are no reasons whatsoever that you should have an alignment of the 3 colors unless you want it so for aesthetic reasons. But experience shows that you'll lose details if you do that.
What I do with narrowband is use the HistogramExpansion to show the best data out of each set independently of the other 2. (I may also clean up the noise and use wavelet processing to maximize the details I want to see) and then, lastly I combine the 3 images using PixelMath.

(see http://pixinsight.com/forum/index.php?topic=2513.msg25601#msg25601)

You can use one color per band or combine them, for example Juan likes this:
0.5*S2 + 0.5*Ha
0.15*Ha + 0.85*O3
O3

[niteman1946]

Hi Steve,
  I followed your guide and was able to align the channels.  Somewhat embarrassed to say I preferred the original color results to the aligned channels.
But thanks nevertheless.


Hi Philip,
  So much to learn, so little time.  Thanks for the link and recommendations on color combinations.

Mark

[sreilly]

I too have not gotten the results I expected with narrow band but then I've only done a few objects and pixel math isn't a tool I'm that familiar with. Looks like I'll need to do some experimenting.

-Steve

[niteman1946]

Thanks again Steve and Philip.

I followed up on both the Hubble and Juan's preferences on color combination.  The Hubble version was just too "Green".  Juan's version introduced some red into the image and its overall look was somewhat salmon (and subdued).
In the end, I prefer my first choice (posted above).  It's more brilliant and highlights the small red nebulosity towards the lower left center.  This was done mostly by accident, although I did favor red to green to blue in descending order.
So as has been oft repeated, the coloration is more a matter of taste.

BTW, Pixelmath is really a powerful tool.

One more question,
I did the pixelmath combination at the start of the process, and then all of the normal steps afterward.  Would I expect better results if I did for each (Ha, O3, S3) the linear stuff separately, and then combined just prior to permanent stretch?

Mark

[Philip de Louraille]

With narrowband data, like I mentioned above, I do the PixelMath combination after I have stretched and processed each channel independently. You'll get more details that way.
Once I combine, then using ColorSaturation, I try to emphasize the colors without getting crazy about it.

[niteman1946]

Should have read before I asked.

Thanks Philip.

[niteman1946]

Philip and Steve,

This is a follow-up using the Hubble pallette.  It came out pretty good, although star color needs work.

Thanks for your help.

Mark

[Philip de Louraille]

The stars's color is because the "sizes" of the stars are different for the three bands.
Good job by the way.

[niteman1946]

Hi Philip,

Thanks for the compliment.

I took the three processed and stretched images (Ha, O3 and S2), just prior to PixelMath combining.  And tried to reduce star size of two (O3 and S2) using Morph Trans to slightly below the Ha.  This was all done by simply comparing star sizes by eye.

Noticeable was that O3 and S2 had substantially more flare surrounding their stars.  Ha was cleaner with little flare.

Nevertheless, after reducing the O3 and  S2, I combined the three and the stars had much less magenta than the previous combine.

Are there more rigid techniques for taming the star colors?

Mark