Isolating a bright star
Posted: Fri Dec 04, 2020 8:49 pm
In my image of the Gamma Cass Nebula (IC 63), I did a lot of work to control the bright star. My technique was to create three separate images from the source Ha luminance layer: one each for nebula, regular stars, and bright star. This worked because the bright star did not overlap the nebula. It would not have worked otherwise. Stretching the nebula in the entire image would have blown out the bright star, because the bright parts of the nebula are brighter than much of the star's halo. Separating the parts was the only way to stretch the nebula without blowing out the star.
Original image. This is a simple STF stretch of theHa stack, so you can see what I was working on.
All the steps below were performed after doing a Masked Stretch of the image.
Step 1 was to isolate the regular stars, using Starnet.
- Clone the image.
- Use Starnet on the clone to create a starless version.
- Use PixelMath to subtract the starless image from the original, leaving a stars-only image.
This is the stars-only image, after some star-reduction with Morphological Transformation.
Sorry, I did not do an intermediate save of the starless image. It has the nebula and most of the bright star, including its halo and diffraction spikes.
Step 2, setting up the extraction:
I then created a PixelMath function that would extract the star only. The expression I used was:
$T*iif(sqrt((x()-2428)^2+(y()-2044)^2)<300,1,1-(sqrt((x()-2428)^2+(y()-2044)^2)-300)/330)
(2428, 2044) are the (x,y) coordinates of the centre of the star. 300 is the radius of the centre part where I wanted 100% extraction. There is a transition zone 330 pixels wide (i.e. total radius = 300+330 = 660 pixels) from 100% extraction to 0% extraction.
This is the result of the extraction, when the above formula was applied to the starless image:
Step 3, performing the extraction:
Subtracting this from the stars-only image (PixelMath again) left a nebula-only image:
(This copy of the image has already been stretched.)
Step 4, stretching the nebula:
The key to avoiding artifacts at the circular "hole" where the star was is to identify where on the histogram the transition occurs between the nebula image and the star image. Here is the Curves control:
Note that "Track View" is selected, so the histogram is displayed on the graph.
Since the peak of the histogram represents the background, the "toe" to the left of the peak represents the part that is darker than the background, namely the "hole" left by the star and its transition to the background. The nebula lies to the right of the peak.
To avoid changing the transition, I pinned levels for the hole and the background by defininging two points on the 45-degree straight line, at the left and right of the peak. I could then raise the rest of the curve, brightening the nebula, without affecting the transition area. The nebula-only image above is the result of this stretch.
I deliberately did not stretch the star at all. That was the point of isolating it from the nebula.
Step 5: recombining:
Because the separations were performed by subtraction, recombining them is a simple PixelMath addition. The result is shown below.
Original image. This is a simple STF stretch of the
All the steps below were performed after doing a Masked Stretch of the image.
Step 1 was to isolate the regular stars, using Starnet.
- Clone the image.
- Use Starnet on the clone to create a starless version.
- Use PixelMath to subtract the starless image from the original, leaving a stars-only image.
This is the stars-only image, after some star-reduction with Morphological Transformation.
Sorry, I did not do an intermediate save of the starless image. It has the nebula and most of the bright star, including its halo and diffraction spikes.
Step 2, setting up the extraction:
I then created a PixelMath function that would extract the star only. The expression I used was:
$T*iif(sqrt((x()-2428)^2+(y()-2044)^2)<300,1,1-(sqrt((x()-2428)^2+(y()-2044)^2)-300)/330)
(2428, 2044) are the (x,y) coordinates of the centre of the star. 300 is the radius of the centre part where I wanted 100% extraction. There is a transition zone 330 pixels wide (i.e. total radius = 300+330 = 660 pixels) from 100% extraction to 0% extraction.
This is the result of the extraction, when the above formula was applied to the starless image:
Step 3, performing the extraction:
Subtracting this from the stars-only image (PixelMath again) left a nebula-only image:
(This copy of the image has already been stretched.)
Step 4, stretching the nebula:
The key to avoiding artifacts at the circular "hole" where the star was is to identify where on the histogram the transition occurs between the nebula image and the star image. Here is the Curves control:
Note that "Track View" is selected, so the histogram is displayed on the graph.
Since the peak of the histogram represents the background, the "toe" to the left of the peak represents the part that is darker than the background, namely the "hole" left by the star and its transition to the background. The nebula lies to the right of the peak.
To avoid changing the transition, I pinned levels for the hole and the background by defininging two points on the 45-degree straight line, at the left and right of the peak. I could then raise the rest of the curve, brightening the nebula, without affecting the transition area. The nebula-only image above is the result of this stretch.
I deliberately did not stretch the star at all. That was the point of isolating it from the nebula.
Step 5: recombining:
Because the separations were performed by subtraction, recombining them is a simple PixelMath addition. The result is shown below.
