Introduction
There are several ways to get to a colour image. The most common is RGB, three channel masters representing the red, green and blue filtered images taken at the camera. Much use is made of a Luminance layer or channel, one that is taken through a clear filter or, sometimes, H alpha. It should be remembered that luminance may help to inject more detail, especially if using binned colour, something I don't recommend, but it invariably dilutes the colour and can give rise to other problems. You can make a false luminance channel from a mix of R, G and B in any proportion you like but there is no new data here so it is not the same as a discreet luminance channel. The channel masters can be combined in Maxim, which to me makes the most sense, or it can be done in Photoshop (PS), not a particularly sensible method. If using a luminance layer doing the mix in PS does give you more flexibility but only if adding the luminance to a completed RGB. Both methods are discussed below. You can make a luminance layer and boost the contrast to help the overall picture as boosting the colour channel contrast risks raising the noise floor but too much contrast too soon in the process can be counter productive.

The previous page described how to go from a raw channel file to a finished mono image. The exact same method is used but with some extra adjustments being made to ensure there is no undue colour bias or cast. Much is made of getting the colour balance right by using a G2V star to calibrate your exposures at the camera. Unfortunately under British skies this does not seem to work too reliably and in any case you are only setting the 'white' colour for stars or anything of equal brightness. If you are guilty of agressive processing and have caused much curve crossing between channels yours stars may end up white but the nebula could be any colour under the rainbow.

I tend to expose my ST10 with SBIG filters using the weighting given by SBIG, 1:1:1.6 but with the STL using Astronomik filters I am currently using 1:1:1. Once you get to adjusting the colour using Curves, Hue/Saturation, Selective Colour and any other device you hear being used the world is your oyster, you can have any colour you want. G2V calibration seems a world away, particularly in view of the idea of 'pretty pictures' rather than science.

Procedure
The three or four channel masters are prepared as normal, ie, calibrated, aligned and combined by your favourite method. They are not stretched in any way. In Maxim they will be displayed with a certain screen stretch but that is all it is, just a stretch for the benefit of the screen display. I do not do anything to these channel masters after they are combined except align them and save the channel alignment and then open them and choose (in Maxim) Colour - Combine Colour.

The resulting file is saved as a 16 bit integer TIFF. At this point there should be no need to go back to Maxim and no need for FITS Lib. Maxim can be closed and PS opened ready to process the colour image.

The second illustration shows the picture after three iterations of Levels. In this case, because the picture had ten hours of exposure it was necessary to raise the black slider slightly to prevent the background from becoming too light. It was only a small move to reset the background to about 30%. This is where, if you move the black slider too far you lose sight of the faint detail. Setting it by looking at the histogram is quite wrong, it is the picture you should be looking at and measuring - not the histogram.

Here you can see the picture just becoming visible. It is at this point that I switch to Curves for the much finer control possible.

The next illustration shows the first application of Curves. This is a standardised kind of procedure but the curve is not in any way a 'standard curve'. The adjustment points on this curve, and all others used, are set by Ctrl clicking in the picture, thus the curve is precisely tailored to the picture. There is no guesswork and no values are ever typed in the box in the curves dialogue.

There are two curves superimposed in this picture. The two lower adjustment points were placed by ctrl clicking in the picture as shown by the white arrows. When the upper point is moved up (by the arrow keys) the curve bends at both bottom and top. Where the curve begins to flatten off is the danger point in any picture, the top one representing saturation of the high values, as shown at 'B', and the lower one shows similar flattening of the low values as shown at 'C'. In this case the lower flattened portion does not matter as the darkest part of the picture is at the lower adjustment point. This is simply taking up space at the left hand end of the histogram but you cannot see it with Curves. The third adjustment point shown at 'A' is essential to stop flattening of the curve at the top as that leads directly to saturated stars which are beyond any further help. This is placed manually, the aim being to pull down the top of the curve as shown without introducing any sharp bends or kinks.

The last picture here shows the second application of Curves using a similar curve to the previous one. The adjustment points were chosen by Ctrl clicking in the picture at the darkest and lightest places of interest. It is important to remember that the steeper part of the curve acts on a region corresponding to that brightness. If you steepen the curve down near the bottom you will increase the contrast in the darker parts of the picture. Usually the most important part of any picture is the mid tones so your contrast increasing should be concentrated there. Beware of getting carried away as it is easy to flatten off the higher brightness regions so they have virtually no contrast at all. Curves is a 'robbing Peter to pay Paul' kind of business and several careful iterations are usually better than one or two big ones.

The bottom of the curve has been flattened again but this is the automatic shrinking of the space at the end of the histogram that often happens with curves adjustments. This is why I never concern myself with space at the end of the histogram when working with Levels. The picture here is looking rather sad due to various gradients which I usually deal with using Quick Mask gradient selections and adjustments in Curves to get the balance and brightness.

The picture shown here is a rollover version of the one above. Mouse over the picture to show the nearly finished version. All the gradient problems are obvious and were taken out by the quick mask method mentioned above. This picture is a narrowband HaS2O3 mapped to RGB, just over ten hours exposure.

The nearly finished version has enough contrast stretching to make a workable picture but not a finished one. I stop stretching at about this point, when I can see most of the detail but the picture is a bit flat. The operations that follow such as Possible layer blending, high pass filtering, unsharp mask sharpening etc can all have an effect on the contrast of the picture. Take it too far at this stage and the contrast increasing effects of sharpening (that's all it is) will slowly begin to destroy it.

This picture is a crop of about 800 pixels wide from an original of 4000 pixels wide so it is suffering from scaling and jpegging. The finished version can be seen here.

In summary, the method outlined above is as simple as I can get it. That doesn't make it foolproof but it helps.

Do all the pre-processing (calibration, aligning and combining) in Maxim, save as a 16 bit integer TIFF.

Open in PS and do three to four iterations of Levels.

Follow with two to three iterations of Curves. Do not try to achieve a finished picture yet.

Attend to any gradients now that you can see them clearly.

Do the other necessary work such as high pass, noise reduction and sharpening.

Finish off with some gentle curves adjustments, crop and add a border if it helps.