In a recent project with my friend @Eric Coles (coles44) , we decided to process the Pillars of Creation data from the James Webb Space Telescope.  My version of the NIRCam data is shown above (look for my version of the MIRI data shortly...I like it better!).  Eric and I thought it would be a good idea to share our procedure for acquiring and processing these data for those who might like to attempt the same, but may be uncertain of how to do so.  I hope you find this helplful.

And, make sure you check out Eric's Version to see his results and read about his method for producing it as well.

Acquiring the data
All data from the JWST are publicly available (as is data from HST, Spitzer, etc.).  To acquire it, simply go to https://mast.stsci.edu/portal/Mashup/Clients/Mast/Portal.html.  At the top, enter the target name, in this case M16, and press search.  When the list is displayed, filter it by selecting JWST in the “Mission” listing on the left.  A list of downloads will be shown.  This image is created using the NIRCam data (identified in the “Instrument” column).  Download the files whose name ends in the following:
    F090W
    F187N
    F200W
    F335M
    F444W
These names are the names (and frequencies) of the narrow band filter used.  In each of these downloads you will find a FITS file with the data.  You now have the data necessary to produce the image.

Processing
I used a combination of PixInsight and Photoshop to process the images.  There are many ways to approach this task.  This was my approach.

    1.    Open the images in PixInsight.  Rename them F090W...F444W as shown above for convenience.
    2.    Use DynamicCrop to crop the outer edges from each image.  Use the same crop for each image (they are already aligned).
    3.    Apply the screen stretch as a permanent stretch using HistogramTransformation to each image.
    4.    Create an artificial luminance image “L” by combining each of the images using the max() PixelMath function.  
    5.    Use whatever typical processing steps for a stretched luminance image you typically use to process “L”.  I used HDRMultiscaleTransform, LocalHistogramEqualization, and MultiscaleLinearTransformation to enhance and sharpen the details of the L image and used Curves to enhance the contrast a bit.
    6.    Export each of the color channel images and the L image as 16 bit tiff files.  We will use photoshop to combine the images.  Note that this could be done using PixelMath, but I find that PhotoShop is a simple and flexible tool for this task.
    7.    Load each of the exported tiff files into layers in PhotoShop.  Place L at the TOP of the list of layers , then each of the color layers below in decreasing frequency, so F444W, F335M, F200W, F187N, F090W.  (Actually, the order of the color layers does not matter, but I find the organization to be helpful).
    8.    Place a solid color adjustment layer at the bottom with a color of black (R:0, G:0, B:0).
    9.    For each of the color layers do the following:
        a.    Add a group with the name of the group as the name of the filter.
        b.    Place the color layer in the group.
        c.    Add a HueSaturation adjustment layer to the group.
        d.    (Optional) add a Curves adjustment layer or a Levels adjustment layer to the group (whichever tool you prefer...I like curves).
        e.    Change the blend mode of the group to “Lighten”
    Each color layer should look like this:


    Others will often omit using adjustment layers for this purpose. I find that using adjustment layers affords me the opportunity to go back and make adjustments.  I think this is the most flexible approach.
    10 .    We will use each Hue/Saturation adjustment layer to colorize each individual color layer.  Check the “Colorize” checkbox and select the desired color from the spectrum.  The colors you choose are certainly a matter of taste.  I chose to break up the color spectrum somewhat evenly between each frequency, assigning the longest wavelength from the visible spectrum to the lowest frequency filter.  So:
    F444W -> red
    F335W -> yellow
    F200W -> green
    F187N -> cyan
    F090W -> purple
    Keep the saturation LOW (I used a value of around 25 to start with for each channel).  Turn off the visibility of the L channel and each of the other color channel groups for this process so that you can easily visualize your changes.  Leave the black solid color adjustment layer at the bottom of the stack visible.
    11.    Make each of the color layer groups visible. If you added them, you can use the individual curves layers to adjust the brightness/contrast of the individual color layer if desired.  You can also adjust the opacity of the group folder to reduce the contribution of a particular layer if you like.
    12.    Change the blend mode of the L layer to “Luminosity”.
    13.    This is the end of our PhotoShop processing.  You can either perform your finishing touches in PhotoShop.  Or, as I did, you can export a flattened tiff image from PhotoShop and return to PixInsight for final processing.  I adjusted curves and saturation a bit, and I flipped the image horizontally, as well as a few other finishing touches.

About Diffraction Spikes
Because of the unique design of the JWST mirror, the diffraction spikes from the stars are unique and very crazy!  If you zoom into the stars in this image you will see what I mean.  Some choose to attempt to process these out.  I attempted to do the same with various methods.  I was successful in doing so, but I felt that the image began to lose something and looked too over processed.  In the end, I chose to leave the diffraction spikes as they are.  In the end, I think the unique look screams JWST and should be a part of the final image.

Please take a look at Eric Coles’ results and read about his processing method, which is slightly different than my approach.  And please let me know if you try this procedure.  I would love to see your results.

Best.
Tim.