Technique for combining Ha and OIII images
I keep this page for historical reference only, since I have modified
this technique as of 12/31/05. The recommended bicolor technique
can be found here.
filters such as those that pass SII, Ha, and OIII light
permit moonlight imaging of supernova remnants, planetary nebulae, and
several diffuse nebulae (such as IC1805, M16,
IC1396, etc.). Combining images from narrowband
filters is often done using the Hubble
tricolor palette, in which SII, Ha, and OIII are assigned to R, G, and
respectively. Another technique adopted by the
Canadian-France-Hawaii Telescope team (CFHT) is the sequence Ha, OIII,
and SII (for R, G, B, respectively). With either of these
techniques, the resulting image is portrayed in "false color" and takes
some getting used to. In addition, star colors are not
well-preserved, and many images using the Hubble palette are
characterized by red halos that are the result of greater bloating of
stars with the SII filter (partly due to the need for increased
stretching of this channel).
I have developed a processing method that only requires data from Ha
and OIII filters, in an attempt to 1) decrease the amount of
imaging time necessary for each target, 2) produce a more "realistic"
looking color image, and 3) still preserve some of the unique
appearance of narrowband imaging. The technique described below
accomplishes these goals by creating a novel
channel ("sG"), by using data from Ha and OIII. I refer to the
final color composite as Ha:sG:OIII. The colors
are pleasing and are representative of Ha (red) and OIII (blue),
ionization fronts are yellow, and the stars do not have halos (although
relatively colorless, which is a characteristic of this technique).
Ha is used for the R channel, and OIII is used for the B channel.
The synthetic green channel is created by multiplying the OIII layer
with the Ha layer. Construction of the color composite is done
using the layer method in Photoshop CS and should be followed exactly
described for best results.
here for another version of this technique which uses the
Hue/Saturation adjustment layer instead of the Color Balance layer.
(please click on the indicated links for an illustration of the steps):
1. Obtain your best Ha and OIII
the usual manner. Process as usual.
2. Layer the images as shown here.
Label as indicated.
3. Change the Mode to RGB (i.e., "Image, Mode, RGB Color").
When prompted, choose "Don't Merge"!
4. Duplicate the Ha and OIII layers
in anticipation of creating the synthetic green channel. Label as indicated.
5. IMPORTANT- Assign "lighten" blend mode to the OIII (blue channel)
assign "multiply" blend mode to the OIII (synthetic
"screen" blend mode to Ha (synthetic green)
6. Create a clipping mask using the OIII (synthetic
green) layer (i.e., click on this layer, then go to "Layer, Create
7. We will now assign an adjustment layer to each color group, in order
to colorize them. First, click on the Ha (red
channel) layer, go to "Layers, New Adjustment Layer, Color
Balance". Be sure to assign this as a clipping mask. You
now have a Color
Adjustment Layer for the Ha (red channel). Label
as indicated. Don't
now- we will do this shortly.
8. Do the same for the other two channels
(note the location of the Color Balance layer for the synthetic green
group!). Label as indicated.
9. Now we are ready to colorize each
channel. This step involves
the sliders of each Color Balance layer. Here is a suggested slider
profile that will serve as a good starting point:
Color Balance (red) layer:
+90, 0, 0
+90, 0, 0
0, 0, -20
Balance (green) layer:
0, +90, +20
Midtones: 0, +90, -70
0, +20, 0
Balance (blue) layer:
0, 0, +30
Midtones: 0, +20, +20
0, +50, +50
10. Addendum (10/29/05)- I have used
this technique quite a bit now and find that these starting values do
not need further adjustment. Rather, additional adjustments at
this point should be made in 3 ways:
A) If there is clipping in the R, G, or B channels,
simply decrease the opacity setting of the respective color adjustment
layer for that channel, until it is corrected.
B) If the histogram shows that the color channels do
not start at the same blackpoint, simply adjust them individually using
a "Levels" adjustment layer so as to achieve a neutral background
(i.e., bring the black point up to the histogram rise for each channel,
so that all colors start at roughly the same point on the histogram
C) The most important adjustments are now done using
the "Selective Color"
layer as a top layer (no clipping mask). You can
also right click next to each layer in order to assign it an
identifying color, which makes it easier to work with. Here's an example of a final
11. You may also choose to flatten this image and use it as color
for the Ha luminance channel, for a final composite of Ha(Ha:sG:OIII).
12. Here is an example of the technique in action, using only the
Ha and OIII channels from a Hubble Palette image of M8 from
These two images should allow you to directly compare the Bicolor
Technique with Hubble Palette processing.
Although there is loss of information when using data from only two
channels as opposed to three, the Ha:sG:OIII
image still captures a lot of detail and has a pleasing quality to it
(although this is subjective). I have also used this technique to
process red and blue broadband filtered images obtained from the
Digitized Sky Survey, as shown here.
The Ha:sG:OIII technique takes some patience to master, but I believe
represents a nice way of achieving a pleasing narrowband image using
data from only two filters (Ha and OIII).
Happy moonlight imaging!