Galaxies are among the most fascinating objects to be photographed, their light emission comes mainly from the stars so it has a continuum spectrum.
For this reason, the primary capture method with CCD cameras is through RGB broadbandfilters ot get true-color images.
Some spiral galaxies, however, are very rich in HII regions, areas in which star formation is particularly active, and emit mainly hydrogen spectrum whose main emission line in the visible, called Hα (H alpha), is in red at 656.28 nm.
A second emission line , called Hβ, is located in the Blue at 486.1 nm. Since the physical phenomenon that generates these two lines is the same, the spatial distribution of the Hβ emission is almost identical to that dell'Hα but less intense (about 20%).
In these cases, a narrow band filter centered at 656.3 nm (commonly called Hα filter) can be very useful to increase the contrast of these areas tha the spiral arms of the galaxy.
Unlike photography of nebulae, when dealing with galaxies, the Hα image can not be used in place of the red one, because with this filter the continuum emission is greatly attenuated and thereforewould be impossible to get the correct color balance of the image.
the data set
The images for this tutorial I have been provided by friends astrophotographers Marco Favuzzi and Giorgio Favini.
Both Giorgio and Marco gave me a set that includes images R, G, B, L and Hα of M81-M82
Before proceeding with processing the images must be calibrated and integrated as usual, still linear and any gradient from light pollution must be removed before you can deal with the next steps.
The method principle for composing Hα with the RGB is based on the following observations:
Supposign that exposure time of the two images are the same, the intensity of line emission coming from the nebulae in the Red and Hα images is substantially the same because both filters allow the red radiation at 656 nm practically undisturbed being monochromatic.
The continuous emission of stars instead is significantly attenuated in the filter Hα because its bandwidth is much smaller than the R filter (typically a few nm against a few tens of nm) so in the Hα image the continuum emission is much dimmer .
Assembling the image Hα with the image R it is possible to isolate the contribution of discrete nebular emission from that of continuum emission of stars.
The composition of the two images is done through the PixInsightprocess PixelMath. If we denote by N the image containing only the contribution of the discrete emission the composition formula is:
N= Hα - Q*(R-med(R))
This formula isa a variation o the original Vicent Peris expression.
The concept behind this formula is very simple: R image, multiplied by an appropriate factor Q lower the one, is subtracted from the Hα image. Q dimms the red image just enough to cancel the contribution of the continuum spectrum in the Hα image leaving the only contribution due to the narrowband emission.
The subtraction of the median value of R using the function med(R) ensures that the median value of Hα is not changed and that there is no loss of data for the phenomenon of clipping.
Q is a constant that depends on several factors including the bandwidth of the filters R and Hα and the respective exposure times of the images. Its value is about:
- Wh is the bandwidth of the Hα filter
- Wr is the bandwidth of the R filter
- Th is the exposure time of the Hα image
- Tr is the exposure time of the R image
In fact, rather than calculating the constant, is much more practical to find it by trial using a preview on the galaxy and changing the value of Q until the continuum emssion component of the image almost disappear.
Once you have the image of nebulae can use it to "pump" the nebulae in RGB image, already calibrated and balanced chromatically, inserting it into the Rchannel and, in part, into the B channel to simulate theHα and Hβ contributions.
If you want you can use a special multiplicative factor to accentuate the visibility of nebulosity.
The first step is to separate the HII regions in the Hα from the issue continues from the continuum emission as described in the previous paragraph.
Open the Hα and Rimagesand create a preview that contains the galaxy on the Hα in order to carry out tests to optimize the value of Q.
Open the process PixelMath and enter the following expression in RGB / K
Ha_Image-Q * (R_Image-med (R_Image))
(where, of course Ha_Image and R_Image must be changed with the names of your images)
and the next in the field "Symbols"
Q = 0.143
Enable the Create New Image check and enter and appropriate Image ID
Apply the process to preview dragging the triangle on the bottom left on the image. if the value of Q is correct the continuum emission should disappear completely (as well as the not saturated stars). if the value is wrong the correction will be poor or excessive.
Change the value of Q accordingly and reapply the process to the preview
Note that the saturated zonesin one of the two images will never be adequately corrected.
Value too low
The continuum emission in the galaxy is still visible.
Value too high
The continuum emission in the galaxy is over correctd.
When you find the optimal value apply the process to the whole image to create the new image containing only the Hαcontribution.
The resulting image may be noisy especially on sky background, it is better to apply a slight noise reduction masking important features to not lose important details.
After the denoise minimize the image: we will need later.
Now we need to create an RGB image from the master light R. G and B and color balance.
Use ChannelCombination selecting the right images and apply the process globally (by pressing the "circle") to generate the RGB image.
Neutralize the sky background: create a preview that contains mostly sky background and neutralize it using the process BackgroundNeutralization; before applying the process verify that the parameter Upper Limit is slightly higher than the average level of the sky background. Now that the backgroundis neutral you can color calibrate: Create a second preview on Galaxy and use it as a white reference (disabling Structure Detection) and the first preview as background reference in the process ColorCalibration.
The lower limit of the white reference and the upper limit of the reference background should be slightly higher than the median value of the sky background.
Now the RGB image is ready for the addition of the Hα contribution . Again we are going to use PixelMath to perform the operation: We use three different expressions for the R, G and B so uncheck the Use Single RGB/K expression and enter the following expressions
R/K: $T+B*(Ha_Clean - med(Ha_Clean))
B: $T+B*0.2*(Ha_Clean - med(Ha_Clean))
H Clean is the Hαimage purified by the contribution from continuum emission; B is the so-called boosting factor which serves to adjust the intensity of the effect: its typical value is between 2 and 4, a higher value leads to an excessive effect, too low a value instead makes the effect imperceptible.
The process must be applied to the image RGB (as always you can use a preview to try different values of B)
As you can see the expressions in the Red and Blue are the same apart from a factor 0.2, in fact, the contribution Hβ is about 20% of 'Hα here is the reason for that factor.
The green channel instead remains unchanged by the operation.
It is important to subtract the median Ha_Clean via the function med(Ha_Clean) so as not to alter the color balance.
In the next image can be seen the resul of the operation with an amplification factor B = 3 (part of the image of Marco Favuzzi)
|Before Hα Boost
|After Hα Boost
Mouseover to see the effect of Hα Boost
Quella che segue invece è parte dell'immagine di Giorgio Favini con un fattore di Amplificazione 4
|Before Hα Boost
|After Hα Boost
Mouseover to see the effect of Hα Boost
You should perform the same operation on the luminance channel if present.
the formula is the same of that in the R/K channel of the RGB image
RGB / K: $ T + B * (Ha_Clean - med (Ha_Clean))
Symbol: B = 1.5
Usually, a smaller amplification factor is enough.
After that the Hα channelhas been integrated in the RGB and eventually in L the processingcontinues as usual.
Here's an example of what you can get after a simple processing.
The image is a Hα - LRGB with B = 3 for RGB and B = 1.5 for L
Image by Marco Favuzzi
Here's another example of processing: Image by Giorgio Favini