Sunday, September 7, 2014

Trying out new things with PixInsight: Noise reduction, removing magenta from stars

One of my experiences with PixInsight has been that it is way too easy to overdue the noise reduction and give images a waxy look. When looking for some advice on noise reduction, I found a comparison of various noise reduction algorithms in PixInsight. It recommends to use TGVDenoise as the most effective tool. And luckily I found a tutorial on how to use TGVDenoise.

I found it surprisingly easy to use TGVDenoise (I love the PixInsight tools that don't have a gazillion parameters...). Here is a closeup of an image of the Pelican Nebula:

Before noise reduction:

After noise reduction:

The noise reduction is noticeable but quite subtle - and some noise is still there. Looking closer at some details:


You can see, that the fine details are still preserved!

Next, I stretched the image and got this:

As usual in narrowband images, the stars have a strong magenta tone. But I found a forum post for a PixelMath expression to remove these:

R: $T[0]
G: iif((CIEL($T)>MIN_BRIGHTNESS)||((min($T[0],$T[2])/(max($T[0],$T[2])>MAGENTA_DEFN))&&(mean($T[0],$T[2])>$T[1])),mean($T[0],$T[2]),$T[1])
B: $T[2]

Now, the stars don't have their magenta color anymore. But also the dark areas of the image on the right have less of a reddish hue. The only problem I saw is that it creates noise around really bright stars:



I asked on the Pixinsight forum if there is a way to avoid this.