How to preserve colourful stars in your images

[MarkS]

I'm sometimes asked how I get such nice star colours in my images.  The answer is that those colours are already there in the original stacked colour image, I just deliberately choose to use image transformations that maintain the original spectral balance. 

The quick answer is that I use the asinh image stretch function in IRIS.  I don't know of any other kind of stretching that similarly preserves spectral balance.

The reason to apply a (non-linear) stretch is that a stacked image will be either 16 bits per channel or 32 bits per channel and will contain an enormous range of pixel intensities.  If you adjust the "brightness" of the image so that stars are displayed without saturating then the faint nebulosity will be too dark to see.  If you uniformly scale the pixel values to make the nebulosity visible then the stars will all be saturated.  So we apply some kind of pixel intensity stretching to "brighten up" the faint nebulosity whilst preventing the bright stars from saturating.  This is where I use the IRIS asinh stretch.  The alternatives found in other packages are log scaling, Photoshop "curves" or other variations. In my opinion these alternatives should all display the following warning: "this function will alter the spectral balance of your stars".

To be more specific, what these alternatives end up doing is to "wash out" the spectral balance of the brighter stars, making them appear more white (and colourless).  So suppose I have 3 stars with a G2V spectral balance in my image - a very dim one, a medium bright one and a very bright but unsaturated one.  These 3 stars all start out having the same proportions of R, G & B because the CCD camera sensor has recorded them accurately.  However, after applying Photoshop curves (or whatever) you'll find the 3 stars end up having different proportions of RGB - the brighter they are, the more white and washed out they become.  I have a very strong opinion about this - it is an effective way of destroying astronomical data!

The usual reason given for this is that the stetch is being done using a colour representation that maintains the subjective uniformity of colours to the human eye.  This might be a fair argument when processing a conventional photographic image but I believe that for an astronomical image, greater weight should be applied to maintaining the spectral balance.  If you must, you can always "make it look right" later on.

Mark

[Carole]

Thanks for that Mark. Must put thinking cap onand see if I can find a way using PS.

Carole

[Mac]

there is also another way of preserving colours and that is to clean them.

If you add an adjustment layer using selective colour (Photoshop)
and then for each of the colours, you remove the appropriate negative colour.

so all you do is slide the negative slider down to -100%

under the Red channel slide the cyan down to -100%

why, well red is made up of yellow and Magenta and the negative colour is cyan.
yellow is yellow and the negative colours are cyan and magenta.

so for each of the colours listed change the sliders to.

Red            Slide the Cyan to -100% 
Yellow        Slide the Cyan to -100% &  Magenta to -100%
Green         Slide the Magenta to -100%   
Cyan          Slide the Yellow to -100%   & Magenta to  -100%
Blue           Slide the Yellow to -100% 
Magenta     Slide the Cyan to -100% & Yellow to -100%     

to adjust the amount of cleaning change the opacity to suit.

Mac.