top of page

Narrowband Processing Hacks (Pixinsight)

Updated: Jul 2, 2023


Recently, I stumbled upon a thread in Cloudy Nights: it was from a user who had exceptionally good narrowband data but he was having a hard time trying to process it. Naturally, I did what most astrophotographers would do: swoop in and attempt to help. Take note that I’m not a professional by any means and like many others, I learnt narrowband processing from the bits and pieces of tips I could find online. He was extremely thankful and of course, encouraged me to create a short clip on Youtube to document my process. I was flattered to say the least, as all this while, I’ve been processing my images via this method but rarely do people comment about my processing techniques because not many understands what goes on behind the scenes in creating Hublble-palette images. Also, most tips & tricks online usually work on clean data-and I mean VERY clean data. What about the masses? Not everyone can afford a $20 grand observatory! A little background: I learnt the basis of this processing technique after watching Don Goldman’s Photoshop tutorial on how Hubble Images are created. While it was informative, the only problem was that his images were pretty darn clean. For me, I shoot under heavy light pollution (Bortle 8 INNER WHITE skies) so my subs aren’t 100% clean all the time.

Note: All narrowband images are expected to be calibrated at this stage i.e they’re your narrowband masters fully calibrated with dark frames, flat frames, bias frames etc.

Step 1: (Unstretched state: Cropping)

(I use DeepSkyStacker to stack my images so this is important)

a) Open all three narrowband masters in PI.

b) DO an auto-stretch (STF) on each narrowband masters

c) Open up (DynamicCrop); drag and draw a box over the area where you intend to crop out

d) DO NOT CLICK THE GREEN TICK JUST YET, instead DRAG & DROP the bottom left triangle icon to the OTHER master frames

e) After you have dragged and dropped the triangle icon to the other frames, FINALLY CLICK ON THE GREEN TICK for the original frame you drew your crop box on.

-----You now have your 3 cropped and framed narrowband masters-----

Step 2: (Unstretched state: Noise Reduction)

Focus on one narrowband channel, minimize the other two. (In this case, I usually focus on H-Alpha first)

a) Duplicate Ha, rename this duplicated image it to MASK_1

b) Open up STF and HistogramTransformation ; click on MASK_1

c) Drag and drop the triangle icon of STF to the bottom of HIstogramTransformation dialogue box

To get something like this

d) Click on the X icon of STF and then the SQUARE icon of HistogramTransformation

This makes sure MASK_1 is stretched and ready to be applied

e) Clip a little bit of the darks for MASK_1, this is for greater contrast between dark areas/backgrounds and faint nebulosity (shift extreme left slider in HistogramTransformation by a little or until you deem fit)

Then click on the square icon as you did before in the previous step

You’re now ready to apply the mask to the main narrowband master!

f) Drag and drop MASK_1 to Ha (until you see this square icon on the left bar of Ha)

You will see a huge patch of red, that’s okay! Just remember this: Red= Protected areas, Non-red= non protected areas

Since we want to do Noise Reduction on the BACKGROUND and not the nebula, we INVERT MASK_1

g) Open up MultiscaleLinearTransform with these parameters, I found these parameters adequate for NR without overdoing it (and making the image look like cotton candy)

NOTE: Depending on your data, these values are to be experimented with.

h) Drag and drop the triangle icon into the main Ha image.

Your final image should be something like this. You can toggle CTRL-Z to see the before and after. You’ll notice that the nebulosity remains untouched while the background has much less noise.

Before After

i) At this stage, REMOVE THE MASK. We will be stretching this narrowband image for the next step

Carry out b) to d) for the new Noise-reduced image.

Repeat a) to i) for the other two narrowband masters

Step 3: (Stretched state: Star Removal)

Important: Take your Stretched Ha master from the end of Step 1 and make a copy of it. Rename it to LUMINANCE. This will be your luminance layer in the final step.

a) Open up Starnet++; drag and drop for all three narrowband images from Step 1

you will get something like this

b) Colour combine all three narrowband images via LRGB Combine (checking only RGB)

What Hubble does is to assign each channel to the respective narrowband data

Red: Sulfur II (SII)

Green: Hydrogen Alpha (Ha)

Blue: Oxygen III (OIII)

You may or may not get an extremely green image BUT that’s expected if it’s the former.

Rename file to RGB_starless

Step 4 (Green Removal) Apply SCNR to the RGB_starless image, this will give the characteristic gold and blue hues

(If there are purples in your image at this stage, Invert the image and apply SCNR once again to the inverted image; Remember to invert your image back to its original state!)

At this point, season the saturation and Lightness of the image under

CurvesTransformation to your taste (as final step will kill off some of the saturation)

Step 5: Sharpening and final denoising

Generate Luminance from RGB_starless

Clip some darks for Luminance via historgram transform (Rename Sharpening_mask)

Apply Sharpening_mask as a mask to RGB_starless (DO NO INVERT)

Apply Unsharpmask on this masked RGB_starless


Final step: Luminance Combination

Combine Luminance layer from Step 3 onto RGB_starless (via LRGBcombination)

You should get an image like this

Season your final image with some curve stretching or saturation pulls. I I prefer less contrast to my image so I ended up with this

That concludes a walkthrough of my narrowband Hubble Palette Processing steps!

Got any requests? Feel free to let me know in the comments below or drop me a message!

848 views0 comments

Recent Posts

See All
bottom of page