Astrophotographer Wil Milan made a time-lapsed video of the 2001 Leonid meteor shower from Arizona. The stunning video shows many, many shooting stars from early in the morning of Nov. 18. Also, partway through, a rare fireball explodes and leaves a trail of dust, all captured on film.

This camera doesn't have any kind of viewfinder or even any way to store the image. Instead it is tethered to a computer (I use a Compaq laptop) on which is loaded software that controls the camera, sequences the exposures, and downloads the data from each exposure to store on disk.

Depending on what resolution the camera is set to (it can operate up to 3072 x 2048 pixels x 16 bits deep) the data files can be up to 12 megabytes per frame. For a night of high-resolution imaging I can easily chew up a gigabyte of disk space just storing the raw data frames. But for this effort I reduced the resolution to just a bit more than standard digital video format, which reduced the total size of all the data files to 260 Mb or so.

(I should add that though these are monochrome cameras, they can take color images, but they do so by taking multiple images, each through a different color filter, which are then assembled into a color frame. Not very practical for, say, baby pictures, but perfectly acceptable for shooting deep-sky objects which are happy to sit still for millions of years.)

For the Leonids video I pointed the camera about 35 degrees high toward the northeast, about halfway between the north star and the Big Dipper. (You can see the north star -- Polaris -- as the bright star at the center left of the frame, while the Big Dipper can be seen rising from below the field of view and traveling upward out of the frame during the 4.5 hours of time covered by the video. The dipper is facing to the left, with the bowl at the top and the handle trailing below it.) I picked this location because from photographing past Leonids I knew this was an area that would get a fair number of meteors passing through it.

For the time-lapse I set up the computer to shoot one frame each minute. The camera and computer require 17 seconds to download a data frame and save it to disk, which left me a maximum of 43 seconds of actual sky exposure each minute.

Once everything was set up, I focused the lens on infinity, stopped it down to f/4 (image is better at f/4 than wide open at f/2.8), started the computer sequence running, and just left it alone. I did have to tape the lens focus ring in place because I've learned that modern autofocus lenses tend to drift out of focus when pointed up in the sky for an extended period of time.

The only glitch that occurred was loss of power.

Because the camera and computer draw a lot of current it's not practical to run them off battery power for very long, so in the field I power the whole setup with a little Honda generator that will happily purr along for more than 12 hours on its little half-gallon tank. Unfortunately after setting up at the site I remembered that I hadn't refilled the generator fuel tank from my last session, but I thought that I had more than half a tank, which should have been enough to run the camera setup for the required duration.

Alas, I was wrong, and about 3/4 of the way through the event the generator unceremoniously sputtered and stopped. Fortunately I'd been paranoid enough to have set up my backup power source, a DC-to-AC inverter that plugs into the cigarette lighter of my SUV. When the generator stopped I rushed over, re-connected everything to the inverter, re-started the camera, and continued the sequence -- but not before a couple of exposures were missed, which is why late in the video there's a little skip in the sequence. Oh well.

The power loss did have one slight positive effect: With the inverter running from the car battery it wouldn't be long before the battery would be dead, so I had to start the car and let it idle for the rest of the night. That meant that I could also run the car heater, which helped keep my 10-year-old son warmer while he was sleeping in the car. But he was so hard asleep that I don't think he noticed :-). (He'd been awake earlier for the peak of the shower, but fell fast asleep right afterward. But he did stay awake long enough to be the required fifth observer when we took a one-minute meteor count at 4:10am.)

Anyway, that's how I acquired the data. Once I got home I processed all the frames (using Mira Pro, a scientific image processing package by Axiom Research), including converting them from the scientific format generated by the camera software (a format known as "FITS," the standard image data format used for astronomy) to JPEG format.

I then used a consumer animation program, GIF Animator, to convert the sequence of JPEGs to to AVI digital video format. All that took a few hours but was finished by Sunday night. The full AVI video file is 300+ megs in size, so the biggest problem I had was finding a way to compress it. I know fairly little about video formats, but by tinkering with GIF Animator I found that the Intel Indeo 5.10 video compression CODEC (whatever that is) crunched it down to 7.2 megs and the quality was still acceptable, so that's what I used, and the result is what I sent out.