Full Io Orbit How-To, Part 3 -- Post

Compiling all the images into a video was the most challenging part of the process.  As I've noted in previous posts, pointing of the camera was both manual and done without an equatorial mount, meaning both Jupiter's position within the picture and the Galilean Moons' orientation around Jupiter changed from frame to frame.

Luckily, I write software for a living and had more than a little training in image processing as a college dork.  Putting the two together, I got the images mostly aligned.  There are some very glaring exceptions.  Almost all of these are due to insisting on taking pictures under non-ideal conditions.  Jupiter and the Galilean Moons are still visible through thin clouds, but either a huge halo emerges around Jupiter (too much light) or the Moons become very dim (not enough light).  Both of these conditions tripped up my automated aligning, forcing me to hand-adjust quite a few frames.  I won't go into too much detail here about the software itself, though I'll post it on Github eventually.

One tip to anyone trying to do the same is don't use generalized computer vision libraries.  Most of them do terribly with backyard astrophotos.  They assume a lot of light and a lot of color and a lot of features.  Until images can fill the frame with the object in question, all of these assumptions are bad for astrophotos.  I converted my images to black/white first (literally black/white, no gray; 1-bit for you computer nerds) to obtain better results than trying to use the generalized computer vision libraries.

I used MATLAB for the image alignment and video rendering.  For the soundtrack, I used some clips I captured on my phone, processed and added to the video in Final Cut Pro.  I considered doing some titling and text overlays, but decided sharing the video was more valuable than perfecting the presentation.  Both MATLAB and Final Cut Pro are pricey options in terms of software.  Octave, Julia and Mathematica are all options to MATLAB, though I've used none and can't attest to the pros/cons of each.  For the rudimentary work I used Final Cut Pro for, iMovie would likely have done just as well.

That's pretty much it.  If anyone has questions or comments, feel free to ask or post in the comments section.

Full Io Orbit How-To, Part 2 -- Photography

Actually taking the photos was, of course, the most rewarding part of this whole process.  In addition to looking around at the rest of the night sky and appreciating just how light-polluted Los Angeles is, I got to see the Galilean Moons actually dance their orbital dance, both minute-by-minute and night-by-night.  Capturing what I saw on camera was the least I could do to illuminate this celestial beauty. Here are some of the nuts and bolts details of how I did it.

How Often to Shoot

I decided rather arbitrarily to take a picture every three minutes.  This worked out well for a few reasons.  First, that amounted to about 850 frames.  At 24 fps, the video would be about 35s long, a nice length.  Further, getting at least one picture every half-degree of Io's orbit seemed like another nice metric to hit.  In hindsight, it was overkill, as Io's orbit only traverses about 120-150 pixels of the image with my camera's setup.  However, what's nice about overkill is things aren't ruined by having one bad frame.  Dropping a single frame when taking them every half-degree is pretty innocuous.  Most importantly for me, however, 3 minutes was about how quickly I could take a picture, open it up on my laptop, zoom in, review and take another picture if I didn't like what I saw.  Now that I've done it 850 times, I could probably get this down to 1 minute (though doing so would be stressful).

Camera Settings

Most cameras come with various automatic settings, things like portrait, landscape, action, etc.  At least with my camera (though I suspect this to be true of most cameras), none of these was useful for this endeavor.  The amount of light assumed by all of these presets far exceeds what was in my typical frame.  I decided to go full manual, which most DSLR's will allow.  There are really only three settings that matter: exposure time, aperture and sensitivity (aka "ISO").  I'll go through each of these separately.

Exposure Time

The longer the exposure, the more light enters the camera and the brighter the image.  Without a tracking mount, there is a maximum exposure time before the rotation of the Earth leads to blurring in the image.  At 300mm focal length, I found this maximum time to be about 1/3 second.  Luckily, Jupiter itself is quite bright, so very short exposures can still capture it.  However, to get the Galilean Moons, longer exposures are required.  They're bright, but not that bright.

Aperture

Aperture is the size of the hole that light is allowed in through the camera to the imager.  For the purposes of focusing at infinity, the only thing aperture affects is how much light gets in.  The larger the aperture (low f-stop), the more light hits the imager.  F-stop affects other things in terrestrial photography, most notably the depth of field.  If you don't care about this, you can skip the next paragraph.

For reasons I'm still not terribly clear on, aperture is specified not by the diameter of the aperture, but by an "f-stop" or "f-number."  The f-stop is the fraction of a focal length that the diameter of the aperture is.  So, in my setup (300mm focal length), an f-stop of f/8 means the aperture is 37.5mm (= 300mm / 8).  In addition to allowing more light, lower f-stops reduce the range of distance from the camera in which things are in focus (known as a "depth of field").  Any given lens will usually be specified by its minimum f-stop, which is the largest the aperture can be.  So a 300mm, f/5.6 lens can open its aperture up to 53.6 mm (= 300mm / 5.6).

Sensitivity

Back when cameras still used film, the ISO number referred to how quickly the chemicals on the film would react to light.  Lower ISO numbers meant a slower reacting film.  This setting on a digital camera is intended to mimic old photochemical film.  For astrophotography, getting brighter pictures is almost always good, so using a high ISO is usually preferred.  However, it should be noted that a high ISO also leads to higher noise in the photograph, as things that appear black to the eye don't to the camera.

Final Settings

After some experimentation, I settled on 1/4s exposures at f/6.3 and 6400 ISO.  These were the settings I used on very clear nights.  A discerning eye will notice a lot of pictures taken on not-so-clear nights.  For those, I would adjust the exposure or sensitivity.

Time

While putting together a video of Io's full orbit was quite both challenging and fun, more than either of these, it was a commitment of time.  Io's orbit is 42.5 hours.  There's simply no way around spending that amount of time taking pictures.  Further, overlap with portions of the orbit that have already been photographed keep increasing.  At first, just going out when convenient to take some pictures was unlikely to find Io in the same place in its orbit around Jupiter as it was in photographs already taken.  Towards the end, it was just the opposite and it became harder and harder to find good times for observing.  Missing an opportunity means waiting 42.5 hours for it to come around again.  Even the willingness to just drop everything when that time comes may be thwarted by the rising and setting of the Sun or Jupiter.  This is all to say nothing of a ground-based astrophotographer's absolute worst enemy, clouds.

Finding good times to observe portions of the orbit that haven't already been photographed was crucial.  There are some nice sites devoted to showing the Galilean Moons' placement in the sky (I used this one a lot).  I found their principal usefulness to be qualitative.  When I really needed to drill down to when the next 3-minute window was, I cooked up my own simple Python script.  I will eventually post it to github and update with a link here.

Full Io Orbit How-To, Part 1 -- Gear

Like a good scientist, I want others to be able to reproduce my results.  In this series of three posts, I'll outline each step I took along the way.  There's plenty of information out there on other ways to do this, but this is what I did.

Gear

Camera

The first thing needed to take pictures of the Galilean Moons is, of course, a camera.  For this video, I used a Canon Rebel T5i, though any DSLR should do just fine.  Canon and Nikon aren't the only options, they seem to be the most popular and, as a result, are likely the most economical for a given level of performance.  I hear CCD cameras are good too, but don't know much about them.

Lens

The next thing to consider is the camera lens.  I used a Canon 75-300mm, f/4-5.6 zoom lens set to 300mm.  For the purposes of astrophotography, a camera lens's focal length correlates to magnification (large focal length means large magnification).  Going past 300mm starts to get very expensive.  My lens was only $250.  The cheapest lens Canon sells with a focal length longer than 300mm goes for $1150.  There are also teleconverters available that extend the focal length at the expense of f-number (ie, relative aperture).  These achieve a longer focal length for a lower price, but may sacrifice auto-focus capability (not generally needed for astrophotography, but can be useful).

Tripod

A tripod is a really nice accessory for astrophotography.  The Galilean Moons are bright enough that relatively short exposures compared to constellations or nebula will work; but the necessary exposure time may be long enough that holding the camera by hand isn't steady enough to get nice, sharp pictures at high magnification.  This was certainly the case for me, so I used a tripod for all of my pictures.  The guy at the camera store sold me on a Mefoto Globetrotter.

Miscellaneous

Though not necessary, having a laptop is really nice.  At the magnification being used, touching the camera introduces a lot of wobble and blurs the pictures.  Though a simple, cheap remote would do the job, I really liked being able to review the pictures I took on a big screen as I was taking them.  Also, it saves me the (minor) hassle of transferring images from the flash drive in the camera to a computer later.  Canon's DSLRs can be controlled through EOS Utility 2.

Lens wipes are a must.  Depending on where you live, either humidity or dust may collect on the camera lens.  It's always good to have some wipes around to clear anything off, especially in astrophotography, where the camera is pointed primarily up most of the time.

Jacket, hat and gloves.  Even in sunny Southern California, nights can get cold.  Staying warm for half an hour or so isn't so bad without warm clothing.  Staying out 2-4 hours, even in what seems like a balmy 60F at first, can get chilly.  And even if your polar bear self can hack it, it's best to have these items along and not need them than the opposite.

Gear I Wish I Had

Over the course of spending 42.5 hours with my gear, there were definitely items I wish I had.

An equatorial tracking mount would have made my life way easier.  Pointing each frame individually by hand meant that the placement of Jupiter in the frame as well as the orientation of the Galilean Moons relative to Jupiter changed slightly with each frame.  An equatorial tracking mount would have at least kept each night's set of photos placed and oriented identically.  As I'll note in my post-production post, framing and orienting all 800+ frames was a real chore.  With an equatorial tracking mount, I would only have had to frame and orient 20 to 25 frames.

No matter how nice a set up you have, you can always wish for better optics, either higher magnification or a larger aperture.  Though I used just a camera and lens for the Galilean Moons video, it will be more economical to get a telescope for further planetary astrophotography.  Saturn's ring gap, for instance, can not be resolved with the setup I used for this video.  The expense of just jumping to a 400mm lens (as noted above, $1150) can buy a pretty nice telescope and mount.  Also, a telescope with a given focal length can achieve a wide range of magnification by using different eyepieces, which are far more economical than ratcheting up camera lens focal length.

Full Io Orbit

In case anyone missed it, Jupiter is super cool.  Not only is Jupiter the King of the Planets, but it is so bright, it hides 4 moons that would otherwise be visible to the unaided eye; and each of these hidden moons is interesting in its own right.  Io is the most volcanically active object in the Solar System.  Europa is a prime candidate for the search for life beyond Earth.  Ganymede is the largest moon.  And Callisto has the oldest surface in the Solar System.

Collectively, these 4 moons are known as the Galilean Moons, named after Galileo, who discovered them in 1610.  The telescope had only just been invented, so it's not surprising that with really quite modest equipment by modern standards, the Galilean Moons can be resolved and imaged.

With that as context, I present the below video.  This was stitched together from several hundred photos taken over many nights at 3 minute intervals.  Taken together, all of the frames show a full orbit of Io around Jupiter.

For now, enjoy the video.  A how-to and further thoughts will be forthcoming.  Note that Io starts in front of Jupiter and emerges about 4s in.

Saturn

Another picture taken with the 400mm, f/5.6 lens attached to a 2x extender.

What can I say about Saturn that Saturn doesn't say for itself?  In this picture, you can see the rings really are rings, though you'd be forgiven if you thought there was only one.  Imaging the gap between the Saturn and its rings is especially cool when the rings are inclined this much from Earth's perspective.

Crescent Venus

I rented a 400mm lens with a 2x extender to see what it could do.  This is what Venus looked like.

I was a bit surprised that the image wasn't sharper.  That may have been my fault.  With the 2x extender, I was forced to manually focus and Venus was small enough to be difficult to tell whether the focus was right.

Nonetheless, the crescent shape of Venus is clear.  This was one of the primary pieces of evidence that led to the downfall of an Earth-centered Universe.  Mercury has phases too, but it's really difficult to image, making Venus one providing the convincing evidence.

Jupiter and the Galilean Moons

I happen to be out in my backyard this weekend during that sliver of time when the sky is dark enough to see Jupiter with the unaided eye, but not so dark that the sky is black yet.  Unlike the moon, which is usually visible during most daylight hours, the planets are difficult to spot against a blue sky except at dawn and dusk.  I managed to snap the following picture that also shows the Galilean moons.

I did cheat just a little.  I took a long exposure, which made the sky bluer in the picture than it looked to the unaided eye.  The long exposure is revealed if you click on the picture to enlarge it.  You'll see the motion blur from the Earth's rotation.  I don't have a tracking mount, so I still get "star trails" when I take long exposures at high magnification.

Backyard Astrophotography -- The Big Dipper

Been pointing my camera skyward at night of late.  There's a lot to it, but I'm sort of starting to get the hang of it.  Here is a series of 3 pictures of the Big Dipper with everything on the camera set the same except exposure time.  I like that you can go from a bunch of dots to what are more or less just the principal stars in the constellation (something you'll see much more clearly if you click on the images to get larger versions).

A 30s exposure

A 20s exposure

A 10s exposure

I should note that these are raw images.  Far clearer images can be obtained with stacking.  From what I understand, that's particularly useful when there is lots of detail in the subject (nebulae, close-ups of planets, etc).