The Green Comet
- This topic has 20 replies, 4 voices, and was last updated 1 year, 2 months ago by
.
-
Doug wrote,
I’ve read that as satellites continue to proliferate in our skies, astronomy is really taking a hit. Have you (Nick and Paul) noticed this in your sky watching?
Paul said,
I’m primariy a visual observer, so the satellites don’t both me so much. But for those like Nick, and professional astronomers, they are more serious–the track fo a satellite across an astrophotograph is a bit like a big nasty scratch across an olf vinyl record. Not good.
As Paul mentioned, for visual observers with a telescope, they aren’t an issue unless they are a bunch of StarLinks just dumped by Elon.
Below is a master image that stacked about 23 hours of images (over 1,000 frames). The satellite tracks have been removed, but one wouldn’t see them anyway because the image is linear — it is kinda like what one would see visually with a telescope. This is the Orion Nebula, an area that has a large population of geosynchronous satellites crossing in winter.
Below is an image by Amir H. Abolfath, which was the NASA Astrophotography Picture of the Day for 1 Jun 2021. The image has been fully processed into a non-linear photograph, but the satellite trails were purposely left in. The final image is 207 minutes of total exposure time (69 images of 3 minutes each).
You can imagine how many tracks I had in my 23 hours of images (1,380 minutes total, or over 6.6 times as much total exposure time as Amir’s). One of the reasons we take many images is because you can’t do much with a single image that has a bunch of airplanes, satellites, wind, etc. messing things up.
Sixty, one-minute exposures stacked will give you the same signal to noise ratio and data as a single 1-hour exposure. With 60, one-minute exposures, you won’t loose much data if you have to throw a couple away.
Jerry mentioned,
although, if they take a bunch of short exposures and then combine them, which is what they do anyway, they can just throw out the satellite tracks
Not in the Orion Constellation — you would have to discard too many frames — and it is wasted time too!
There is a better way.
When we stack many images, we are doing more than just stacking, which we called “Integration.”
Integration
A high level explanation of the Orion Nebula picture I took a couple years ago: I stacked over 1,000 images, each with 16.2 million pixels. The following explanation doesn’t cover everything that is done during this “integration.”
The software analyzed the stars in each individual picture and identified the same brightest 500 stars in each, and then stacked all of them using the center of each matching star to perfectly align each image with all the others — thus every single pixel is aligned to the same pixel location on each image.
It gets even more difficult, I purposely did not have the scope pointed to the same exact spot. After each image was taken, I purposely had the mount randomly move a few pixels, so no patterns developed due to any bad pixels in the camera sensor (every camera has numerous bad pixels)
So. . .
16.2 Million pixels X 1,000 images = 16.2 billion pixels the software has to keep track of! Not only does it use a lot of processing power, but it used 250GB of temporary disc swap file storage space.
Using an algorithm/method the software determines values for each pixel with mean values, standard deviations and other statistical stuff that would explode my brain — the end result is it rejects outlier pixels, which can be satellite trails, airplane trails, or something else. So instead of 1,000 almost exact pixels for each singe pixel location, we might end up with 998 pixels for some locations and 1,000 in other locations. The statistically bad pixels are rejected.
This process took several hours for the computer to do.
The final integration file, which I posted earlier, was a single linear 338MB file. Linear meaning the values of each pixel remains as the camera recorded it. I then used post-processing software to change it to nonlinear, which is what us humans can see and might be described as “stretching the color tones.” Below is a nonlinear version.
For those interested or who love math and calculus, here’s a link to reference documentation for the primary software I use and this is just the Image Integration Documentation. Once you open it, use your “search” function to find “Sigma Clipping” and “Winsorized Sigma Clipping” in the reference guide. I use “Winsorized Sigma Clipping” on all my image integrations, which is what “removes” satellite trails.
Clouds and a full moon aren’t making this any easier.
After a lot of time searching and a dab of patience, I came up with this last night — the Green Comet !!
Since my first comet image above didn’t work out well and looked like a green can, I took another stab at it.
Here’s the challenge,
- I need to take time exposures (at least 30 seconds each) and stack the sub frames to see anything
- The mount needs to be tracking something. . .
- The comet moves faster than the stars
- If I track the stars, the comet will be blurred and funky
- If I track the comet, all the stars will have star trails
Using some new-to-me processing techniques, I stacked sixty, 30-second exposures. I used a wide field telescope and camera to provide some perspective of the comet compared to the star field. I had planned on following up with a longer focal length telescope scope (higher magnification) to isolate the comet itself, but clouds rolled in the past couple of days and apparently will be here for most of the next seven days.
If the sky cooperates soon, I make take some more images with a larger scope and a different camera. However I’m not highly motivated. It is more fun to look at it visually with my largest telescope. Processing pictures of comets is time consuming to avoid star trails and also capture the comet without blurring — the effort just doesn’t excite me since I had planned on working on other targets this month.
A cropped image. . .
The bright star is Hassaleh/Kabalinan
From StarFacts.com (click on the link to learn more),
“Hassaleh, Iota Aurigae (ι Aur), is an orange bright giant star located in the constellation Auriga, the Charioteer. With an apparent magnitude of 2.69, it is the fourth brightest star in Auriga, after Capella, Menkalinan and Mahasim. Hassaleh lies at an approximate distance of 490 light years from Earth. It serves as a spectral standard for its class.”
Jerry, that’s a great question and you’re very observant.
Actually it is a giant star to begin with. Of course visual star size depends on its distance from earth and its actual size.
The other thing is, I was tracking the comet, keeping it in the center of each successive picture. If we place a camera on a tripod, 15 seconds is about the longest exposure we can take before the earth’s rotation starts to affect the roundness of the stars. I was taking 30 second exposures, and the mount was moving with the earth’s rotation, but at the speed of the comet’s apparent movement and not the stars.
What all of this means is the stars are slightly oval if you really zoom in, which will make them appear slightly larger.
Image below is a single picture with no processing. It is in a linear stage.
In the same image, below, I have “stretched” the color tones
Stars that we can see through a lens are affected by the quality of the optics and the earth’s atmosphere, and can make stars appear larger. This is called Point Spread Function and is an “abnormal” state. Link to Wikipedia on PSF.
In this one, below, I have increased the color saturation of a single frame to bring out some color. Increasing saturation is a balancing act — bring out some color and not make stars too big or artificial. Saturation does increase the apparent size. We don’t see the pinpoints of light but an Airy Disk (click for link).
If you really zoom in on the above picture, you will see some single (one square) pixels that are blue or red. These are not stars, but are bad pixels. Even a high end $10K camera has bad pixels. When I process a stack of images, the software removes the bad pixels and replaces them with an adjoining good pixel.
- You must be logged in to reply to this topic.
