Ha Combine Help -- Continuum Map Question

[andyschlei]

I recently imaged the Bubble Nebula and the nearby open cluster in Ha, L, and RGB.  I have 90 minutes Ha, 30 minutes L, and 20 minutes each RGB (RGB binned 2x2).

My processing approach was to combine the luminance and Ha (in Pixel Math, L + 3*Ha turned out best), enhance that image (Curves, HDR wavelet, and A Trous wavelet) to make a master luminance frame.  I then added the Ha to the red, ultimately doing an LRGB combine with the data from each color.

The result had the typical washed out red look.  I have done my best to fix it with color balance changes, but the red still really isn't red.  A copy is attached and you can see the full sized image in my web gallery.

I started to try the "A New Approach to Combination of Broadband and Narrowband Data" method from the tutorial, with the intent of making a better red frame to drive the red color.  But when I tried to create the continuum map, I get a very strange result.  The center of bright areas is dark.  If I have used an integer upsample for the red, some pixels are at zero.  With a smoother upsample, they are close but not at zero.

To create the continuum map, I used pixel math (Red / Ha) and rescaled the result.  I have a attached a small image showing the artifacts in the image.

Is this due to starting from a 2x2 and trying to build a map from an upscaled image?  Are there any normalizations I should try before building the map (I am starting from the raw combined data to create the map).

Any suggestion would be appreciated.

[vicent_peris]

Hi,

I think what happens is that your H-alpha images are much sharper, so you have stronger peaks in the center of the stars. What happens if you blur a bit the H-alpha image before doing the division??


Vicent.

[andyschlei]

Vincent,

That makes sense.  I had noted some very bright (0.995 K) pixels in the Ha image and tried just multiplying the entire image by 0.9 to reduce the peak but that didn't work.

What is the correct tool and procedure to blur the Ha image in PixInsight?  My guess is Morphological Transformation, using a median operator.  Is that correct?

Thanks,

--Andy

[vicent_peris]

Until we have the gaussian blur tool, you can blur the image using the À Trous Wavelet Transform tool. Try to disable some of the first layers. Also try to change the scaling function.


Vicent.

[andyschlei]

Thanks, I'll try that.

Now here is what I got from my morphological transformation blurred continuum map when I applied the default GREYCstoration noise reduction to it.  Somewhat pretty, but somewhat bizarre too.  Looks like I should try another tool for noise reduction 

[andyschlei]

Until we have the gaussian blur tool, you can blur the image using the À Trous Wavelet Transform tool. Try to disable some of the first layers. Also try to change the scaling function.


Vicent.

One other question.  When you say scaling function, do you mean the dynamic range extension on the wavelets or the rescale parameters on the pixel math divide?

Thanks,

--Andy

[vicent_peris]

Until we have the gaussian blur tool, you can blur the image using the À Trous Wavelet Transform tool. Try to disable some of the first layers. Also try to change the scaling function.


Vicent.

One other question.  When you say scaling function, do you mean the dynamic range extension on the wavelets or the rescale parameters on the pixel math divide?

Thanks,

--Andy

No, I refer to the parameter in the ATW tool. It's just above the wavelet layers.


Vicent.

[andyschlei]

Thanks for the help.  I think my color data is too poor to make this work.  I've got some better data of the Heart nebula I will try.

And in the future I will not bin my color data when shooting narrowband!