My #1 favorite way to spend vacation time is cycle touring with friends, #2 is having a week to dig deep on a side project that I don’t normally have time for. Last week was the later, I was in Cozumel Mexico for a week with friends and I decided to use my downtime to level up my astrophotography. To make the most of it I bought myself a long-overdue quality tripod, borrowed a star tracker from a coworker, and rented the biggest lens I could find.
I’ve taken my fair share of star photos before. In college a friend and I attempted a homemade star tracker but we were never able to make anything too sharp from it. Hand turning a screw produces about as much vibration as you would expect.
After that I never succeeded in moving beyond finding a dark sky and shooting 30 second exposures of the milky way on a wide lens. Telephoto astrophotography is a whole different story, this story.
* red headlamp
* remote shutter trigger
* iPhone w/ star finding apps
* external battery packs
* electrical tape (this saved the day, unsurprisingly)
M83 (and some trig)
I picked the M83 galaxy because as far I could tell it was the largest galaxy visible to me at this time of year. The ‘M’ comes from Charles Messier, an 18th century astronomer who cataloged diffuse celestial objects like galaxies, nebulas and star clusters.
Before this project I had no intuitive grasp of how large most galaxies were in the sky or if the 400mm lens would be sufficient to see anything. Wikipedia tells us M83 is 40,000 light years wide but the most relevant question for astrophotography is:
How many pixels wide will M83 appear when projected onto a 36mm/4256 x 2832 pixel sensor with a 400mm lens?
This will tell us if it’s even possible to photograph M83 with this camera. A one pixel wide galaxy photo wouldn’t be very exciting. The answer can be worked out with trig and it’s kind of fun so let’s do it.
First let’s work out what the field of view of my lens/camera is. Field of view is how much of the world, measured as an angle, an observer can see at a given time. Horizontal human binocular vision has about a 200 degree field of view.
The relationship between sensor size, lens focal length, and field of view in a camera is:
The 400mm telephoto lens is a relatively long focal length and a 5 degree field of view is quite “zoomed in”. For reference iPhone cameras have about a 60 degree field of view.
Arc Seconds are a common unit in astronomy for small angular values, and are defined as degrees x 3600:
5.15 degrees * 3600 = 18,551 arc seconds
So we know that 18,551 arc seconds of the sky are going to be projected onto an image on my sensor which contains 4256 pixels horizontally. That means each pixel will be responsible for gathering light from an area 4.5 arc seconds wide.
Now we need to figure out how wide M83 appears in the sky.
M83 is 550 arc seconds wide when viewed from earth! We already figured out that each pixel in my camera is going to image 4.5 arc seconds.
M83 pixel width = 500/4.5 = 122
The galaxy will be 122 pixels wide in my pictures, better than nothing!
I still need to talk about star tracking but you stuck with me through the high school geometry refresher and you deserve to see a galaxy now.
This is the first frame where I was able to spot a real live galaxy, M83, the hazy spot in the center.
I didn’t realize at the time how out of focus it was, I wasn’t actually sure what an in-focus star was supposed to look like so I went ahead and shot a stack of about 25 of these with 45 second exposures.
Left: a single exposure
Right: 25 stacked exposures with some sensor calibration techniques applied to reduce noise.
You can see the arms! This was pretty exciting.
Depending on what you measure, it’s between 91 and 147 pixels across. Looks like our 122 pixel estimate was reasonable.
Stacking and Tracking
Galaxies are dim objects and achieving a good signal to noise ratio means exposing the image for tens of minutes or even hours.
The earth is spinning about an axis that is roughly aligned with the north star, Polaris. This causes the stars to move noticeably during long exposures.
With wide angle lenses, exposures of 30 seconds are possible without too much noticeable star movement. With a 400mm telephoto the stars will visibly move in less than 2 seconds.
In order to get our minutes or hours long exposures without star movement we can use two techniques.
- programmatically stack many short exposures
- mount the camera on a device that rotates with the earth, a star tracker.
If your stacked exposures are too short, you run the risk of the fixed noise added by the sensor electronics for each exposure being brighter than your image, rendering it useless or very noisy.
If the star tracker is not very accurately aligned with the earth’s axis, it will not entirely reduce start movement. This places an upper limit on how long you can expose before the star movement appears again.
Straddling these two constraints eventually led me to stacking many 20 second exposures for the final shot. This exposed each shot as much as possible before the star movement appeared due to tracker misalignment.
Autofocus lenses are really hard to manually focus. Sure they have the manual focus ring but unlike manual only lenses the ring is much less precise. It doesn’t feel like it was ever actually intended to be used. I tried a few focusing techniques like changing the focus during a long exposure and looking at which part of the star trail is the thinest. In the end the winning technique was simply to pre-focus on the brightest thing in the sky, Jupiter, before composing my galactic shot.
After I improved the focus I noticed the stars were blurring in my shots again. The blurring was sporadic which led me to believe that it wasn’t star tracking misalignment but slippage in some of the tripod joints. My camera/lens were exerting quite a bit of leverage on the ball head of my tripod it seemed like no matter how much I tightened it I saw movement.
All the movement meant I was only able to get 4 worthwhile 45 second exposures to stack. The second attempt was certainly more focused than the first night, but also much noisier with only 3 min worth of integration.
I slept on it and woke up with an idea. I taped about 2 pounds worth of external battery packs to the end of the camera lens as a counter weight and it made all the difference. The tripod stopped slipping around and I was able to see that the remaining error was probably due to star tracker misalignment. I had none of the tools needed to do an accurate polar axis alignment of my tracker, the only thing I could do about that was reduce my exposures down to 20 seconds.
Finally, I was able to get a consistant sharp series of 47 20 second exposures, which comes out to 15 minutes of stacked exposure.
Using all the tricks I learned with M83 I snagged this stacked 20 min exposure of the Orion Nebula on the last night in Mexico. The Nebula is 10 times the angular width of M83 and one of the brightest and largest deep space objects after Andromeda Galaxy. You can see it with your own eyes, just below Orion’s Belt.
I never knew a nebula this colorful was accessible to amateurs. Looking back on my old photos though, the Orion Nebula has been right in front of my face the whole time (see wide shot at the beginning of this post).
I didn’t spend too much time talking about the star tracker in this post but the one I borrowed, Vixen Polarie, is basically a “pocket star tracker”. It was unstable with a camera/lens the size of mine and didn’t come with any calibration scope for accurate alignment. I’m going to be looking into a more serious tracker and also potentially my own telephoto prime lens.
I also want to hit up some star parties to get a feel for what it’s like to image with a telescope instead of a camera lens.
This post obviously isn’t a comprehensive guide. I skipped over the entire image calibration and post processing process which took about as much time to figure out as the image acquisition did. I highly highly recommend Thierry Legault’s Astrophotography if you are interested in getting started.