In the climactic gun battle of John Woo´s deliriously nutty identity-switch action film Face/Off, the two enemies-Cage and Travolta-and their sidekicks fire away at each other with Berettas and Glocks; you can see densely detailed plumes of smoke erupt from the muzzles, obscuring the shooters´ hands. As bullets emerge from weapons, thousand-foot-per-second velocities slow to a graceful glide, until the spin from the spiral rifling of the gun barrels
is clearly visible. It looks like a beautiful bit of computer-generated-image graphics-less fancy than the ripple-wake, stop-dead bullet magic in the new Matrix movies, but somehow more real.
That’s because the Woo images are more real than CGI: Real bullets were slowed on film by a one-off, hand-built device called the Millisecond Camera, shooting 12,000 frames per second. For all the advances in digital effects, 35mm film still offers more information for the eye: “When you really look at it, you can tell it’s film,” says Nathan Nebeker. “You just don’t get that richness of detail with CGI.” Nebeker knows what ‘s talking about. He is the owner of a company called Conniption Films, and he made the Millisecond Camera.
Nebeker’s camera bears scant resemblance to standard movie cameras. “You couldn’t make a mechanical shutter open and close quickly enough,” he says. “[And] there are physical limits to how fast you move film through a camera. To go really fast, you have to move the light to the film.” In his camera, a loop of film rotates rapidly in a drum, whose housing is emptied of air to reduce friction. A rotating mirror and a series of optics directs a “slice” of light to each frame. At speed, the camera holds only enough film for a hundredth of a second’s worth of action, but that’s not quite the problem it sounds: A bullet flies 10 feet in that time.
But the most extraordinary thing about the Millisecond Camera may be this: In Nathan Nebeker’s world, 12,000 frames per second is an excruciatingly slow clip. Running Conniption Films is Nebeker’s second job, a Hollywood- and TV-focused spinoff of the family business: building really-high-speed movie cameras. Cordin Scientific Imaging, based in Salt Lake City and headed by Nathan Nebeker’s father, Sid, has been in the sometimes secretive service of the scientific and military community for almost 50 years. The fastest cine camera produced by Cordin today is unimaginably quick: One second of images taken with the 200-million-frames-per-second digital model would take 96 days to view if played back at standard movie speed.
Picturing the Bomb
High-speed still photography dates back decades, most famously to the spectacular milk-drop and bullet-through-apple photography of Harold Edgerton, who invented the strobe in 1931. Very-high-speed movie photography started later, and came into its own as a scientific tool in the birthing room of the atomic bomb. Its purpose has been not so much to freeze a moment in time, but to glimpse what changes from one infinitesimal moment to the next. It begins at the point at which our familiarity with ordinary time-splitting ends, a point not far past the 100 milliseconds that constitute the blink of an eye. Ordinary cameras freeze time, but crudely and in big chunks. Movie cameras fool the eye into seeing continuous motion when only 24 still frames are presented per second. We know from snapshot experience that the 1/500th or 1/1,000th shutter setting will stop most human motion. Beyond this realm–into micro- and nano- dimensions of time, let alone across exotic picosecond and femtosecond frontiers–the very fast is almost as invisible to us as the extremely small. The eye does not register, and the conventional camera does not record.
High-speed movie cameras do record, but many of the most arresting picture sequences taken with such equipment have not been seen–except by a few researchers who might go to jail if they released them to the public. Secret weapons research is in the DNA of this business. During the final stages of the Manhattan Project, Los Alamos scientists hit a brick wall. Sid Nebeker recalls hearing an account of this little-known chapter in history from Berlyn Brixner, a technician who still lives quietly, at the age of 96, in Los Alamos. The scientists who were developing the theoretical and practical framework for implosion weapons hadn’t been able to get them to work properly. There was much disagreement about whether that failure meant that the entire concept was flawed, or merely that the execution needed adjustment–perhaps, for example, the shape of the explosive “lens” that was supposed to trigger the nuclear reaction needed to be modified. Brixner was enlisted to capture the footage that would allow the scientists to see what was happening.
The camera Brixner used was an early rotating-mirror design. A mirror in the center of a cylindrical housing projected images in series from the camera’s main lens to the film, which was seated along the inside edge of the housing. The light traveling to each frame passed through its own set of lenses en route. The mirror acted as a sort of shutter, flashing a discrete image onto the film and forming a discrete frame. This design had been adapted from work done by C. David Miller, an engineer who worked at NASA’s predecessor, NACA. “With Miller’s concept of forming the image on a rotating mirror and putting a sequence of lenses between mirror and film, you can jump to 1, 2 or 5 million images per second,” says Sid Nebeker. “Extremely crisp ones at that.”
There was room only for a short bit of film, of course: The camera could shoot only 24 frames. But you could time exposure within a tenth of a microsecond, an interval so precise that when the test failed again, scientists had a frame-by-frame record of the event.
“Brixner’s camera was like a ray of light into a very dark situation,” says Nebeker. The problem indeed lay with the uneven detonation of the conventional explosives. A movie camera cleared the final serious obstacle in the path to the atomic age.
The original Cordin camera, built in 1956 shortly after the Manhattan Project technology was declassified, and capable of 1.25 million frames per second, was based on the same rotating-mirror design–a design that is still at the heart of the company’s biggest, most expensive machines.
High-speed camera manufacture has always been a small, clannish industry serving a highly specialized academic community-shock-wave and cavitation physicists, hypersonic-aviation researchers, materials scientists interested in the dynamics of fractures, cracks and vibration–as well as the military research labs working on “energetic materials,” ballistics, bombardment and bombs. The most recent biennial conference of the trade’s worldwide professional organization, the International Congress on High Speed Photography and Photonics (which includes all forms of high-speed imaging, still and moving), had just 226 attendees. Much work involving high-speed cameras continues to come out of labs such as Los Alamos and Lawrence Livermore, with their deep roots in military-tech research. And much of the photography continues to be secret–not only for security reasons but because a car company studying, say, combustion efficiency in an engine fitted with transparent cylinder heads keeps competitive data to itself. “I think I saw some stills [of weapons research] once in a slide show at a conference,” Nathan Nebeker says. “That’s as close as I’ve gotten to seeing the footage.”
The international market is constrained as well: Cordin cameras must be licensed for export by the Department of Energy, which regulates technology related to nuclear weapons. In 1990, an import-export firm located in New Jersey approached Cordin to try to buy one of its cameras for more than $200,000. The end user turned out to be Al Kindi General Establishment, an Iraqi weapons lab involved in nuclear research. Export license denied. “After the first Gulf War,” Sid Nebeker relates, “they found [in Iraq] some Japanese cameras that would have been somewhat helpful. But they wouldn’t have provided nearly the crispness or speed of ours.”
**Hand Built From Scratch **
Sid Nebeker has been running the Cordin company since the early ’60s, when he bought it from its founders, both of whom had been his classmates at the University of Utah. Cordin’s headquarters are located in an industrial section of Salt Lake City, far removed from the gleaming downtown familiar from Olympics coverage. Cordin, which today has about 30 employees, resides in a large, low warehouse that’s divided into an area of spare offices and a sprawling workshop. The office decor is unrelieved engineer-drab: wood-grain paneling, linoleum flooring and, in some rooms, ancient wall-to-wall carpeting.
I found the senior Nebeker in the engineers’ work space, standing, sleeves rolled up, beside a young engineer at a workbench, the two of them tinkering with the circuitry in a camera headed for a South Korean defense lab. Nebeker, 73, is good-looking in a jaunty, square-jawed way, the patriarch of a company referred to as DadCo by his children, three of whom work at Cordin.
Tinkering was a childhood passion. Sid grew up during the Depression on a 20,000-acre sheep and cattle ranch in northern Utah where “the big thing was horses.” Young Sid, however, was more interested in the old machines. At the age of eight he rebuilt a broken-down 1-cylinder engine from a hay gang and used it to power a car.
Ranch life was not for him: He left to study engineering at the University of Utah, spent time in the Air Force, attended Harvard Business School, then returned to Salt Lake City in 1958 to look for work. “There were limited opportunities in Utah for a business-engineering major,” he says. Nebeker spent six months at a fledgling technology company that was going nowhere. In 1959 he met an engineering school classmate named Earl Pound. Pound was part of the faculty at Utah State University and had formed Cordin a couple of years earlier, when the Manhattan Project camera technology was declassified. Cordin had made just one product, the 1956 camera that the Navy bought for use at a weapons facility in Maryland. Since then, Cordin had been dormant: no employees, no customers, no business plan.
Nebeker proposed that the company be re-started, then spent several months at Cordin without pay. He perfected the original camera design, making it “extremely reliable and extremely accurate.” Several months later an order came through from the China Lake Naval Weapons Center, in California. Weapons development is, within the club, a word-of-mouth business, and the camera, along with the China Lake technician who operated it, “became our sales force for the next four or five years. People from the defense industry would call to inquire about our cameras, and we’d say, ‘Talk to Roland Gallup at China Lake. He’s got one.’ He was a marvelous photographer. They’d talk to him and he’d show them some of his work and they’d be sold.” (China Lake still uses Cordin cameras.)
Cordin was off and running in a small niche of the high-speed photography business, though it turned out not to be a very high-speed business; the company today sells about 10 cameras a year, often customizing for specific jobs.
The chief customizer is James Brimhall, 70, who has been with Cordin for almost 20 years and is in charge of camera assembly. The work is as much about handcrafting as engineering. Cordin’s own shop fashions practically every knob, gear and fitting for each camera. With his shop apron, hat, glasses and two hearing aids, Brimhall has an old-world, master-craftsman mien, and is now passing on his expertise and responsibilities to a 32-year-old employee named Lane Oberg. For Oberg, it’s a matter of learning by doing: One does not arrive from optical-engineering school ready to build a device as specialized as a Cordin camera.
One morning I help Oberg assemble one of 160 lens housings that will form the guts of a $385,000 rotating-mirror camera destined for a ballistics range at Los Alamos National Lab. It will be the fifth Model 140 camera the company has built in its history: 2.25 million fps, f/16 aperture, 80 frames total. Its mirror is a pentagonal lump of beryllium coated in polished aluminum; the beryllium has been X-rayed to ensure that the crystalline structure is pristine, lest it explode when rotated at 562,500 rpm by the camera’s helium-driven turbine.
Oberg greets me from his desk, where he’s holding a black, anodized-aluminum mount in one hand and an electric Dremel tool-the kind you see advertised on late-night TV in the other. Metal shavings are piled on the floor nearby. “I knew there was a reason I kept this thing around!” Oberg says cheerfully. “Never used it before, but I figured it’d come in handy.”
He then begins grinding away at parts of the mount, exposing the silvery metal on several surfaces. He explains that the hundreds of lenses that go into the camera will be held in place by these mounts, which are precisely engineered to strict tolerances. But the cement he’s using to hold the lenses in the mounts doesn’t stick well to anodized aluminum. So he’s simply grinding the anodization away.
We duck through a large, black plastic sheet that acts as a UV-light barrier and into what is an almost “clean” room. A plywood board holds hundreds of rectangular glass lenses, which look like superthick lenses for reading spectacles: about 1-1/2 inches long, 1/4 inch thick, maybe 1/4 inch high. Each assembly will hold four lenses, two assemblies per frame, for the 80-frame total. Two of the lenses in each assembly need to be placed with a tolerance in the micrometer range; two do not.
Oberg lets me place and glue two of the lower-precision-fit lenses into an assembly that´s clamped to the workbench. Placing them turns out to be easy enough, gluing them even easier. Just squirt a dollop of cement through an 18-gauge needle, then shine a high-intensity ultraviolet spotlight on it for a few seconds–roughly the same way a dentist cures an epoxy filling.
When Micrometeorites Hit
At 5,000 frames per second you can see a place kicker’s foot connecting with a football as the pigskin slowly wraps around his instep. “To see a golf ball,” says Nathan Nebeker, “you’re up to 12,000 frames per second,” because the club is accelerated to terrific speeds by the lever action of the swing. The next order of magnitude, hundreds of thousands of frames per second, allows you to observe fairly high-speed events: Toyota used a rotating-drum Cordin camera capable of speeds of up to 200,000 frames per second
to peer into its see-through engine. The French navy bought a similar camera, along with a special water-sealed periscope designed and built
by Cordin, to observe the effects of explosives striking ships’ hulls.
But when an object strikes a barrier at 16,000 mph, you need sequences shot in the million-fps range to study the violent effects: Hence the Cordin camera at NASA’s Hypervelocity Impact Test Facility in White Sands, New Mexico, where scientists blast space-vehicle components with tiny projectiles to simulate what happens when they are hit by fast-moving space debris (bits of satellite hardware that have come loose, paint chips) and micrometeorites.
About 100 times a year, a two-stage light-gas piston gun at HITF accelerates plastics and metals in various shapes and sizes (up to an inch in diameter) to 16,000 mph, then smashes them into mock-ups of, for example, the outer skin of a spacecraft. Every exposed part is vulnerable, says Dave Baker, HITF project manager, whether it be “a new particle-impact shield, or a cable, or a tether or anything like that.” NASA has many gas-gun facilities around the country but likes an isolated building at White Sands best. “We can shoot hazardous targets here,” says Baker. “Propellant tanks or oxygen tanks, which might explode, or beryllium, which could be toxic.” Every test is filmed with a Cordin rotating-mirror camera. The footage and other data are analyzed by scientists at Johnson Space Center. Baker was mum on whether the recent space shuttle disaster investigation involved HITF work, though it almost assuredly did: Space debris was one of the possible causes linked to the demise of Columbia.
Stopping Light Cold
The cameras that for the past six years have made up Cordin’s fastest group don’t have moving parts at all–or film, for that matter. These cameras shoot at rates of up to 200 million frames per second, a rate useful, Nathan Nebeker says, for “things like the moment of combustion of aerosolized jet fuel: You’re trying to get an even wave front so that you have a controlled burn.”
In the so-called gated, intensified CCD cameras, light from the objective lens passes through a beam splitter, and multiple, identical images are formed on the photo cathodes of microchannel plate (MCP) image intensifier tubes, similar to those used in night vision goggles. Striking the front side of the intensifier, the photons free electrons inside each tube; the electrons cascade through the tube, freeing more as they proceed, until they strike a phosphorescent surface in back, which converts the electrons back into a light pattern. “The intensifier does two things,” explains Sid Nebeker. “It magnifies the light several thousand times, and it switches the light very, very fast–in a matter of nanoseconds. When you have such a short exposure you need that high light gain to get useful images.”
The images from the MCPs pass through bundles of glass fibers, stretched long and thin at one end to shrink down the picture. The small ends sit against 1-megapixel charge-coupled devices–pretty much what you’d find in a digital consumer camera–which record the images.
Popular Science asked Nathan Nebeker if he could set up a demonstration that would make nanosecond durations comprehensible in visual and intuitive terms. Why not, he suggested, shoot a series of photos of light as it travels in real time across a very short space–say, 100 feet? The fastest thing in the universe stopped dead as it makes its way across an ordinary room.
It is the first time Nebeker has attempted this, though not the first time it has been done–a single photo of light “stopped” by a different technology in the late 1940s was termed “one of the most important photographs of the century” in this magazine (for details, see www.popsci.com/exclusive). “It’s essentially a gee-whiz shot,” Nebeker admits, but not without its challenges. Light travels about a foot per nanosecond, so a 7-nanosecond laser pulse, if you were able to see it, would look like a discrete packet of photons 7 feet long.
Nebeker sets up shop one afternoon at the labs of Spectra-Physics in
Mountain View, California, where he can use a laser capable of generating
7-nanosecond pulses of brilliant green laser light. The camera he brought shoots only two frames; Nebeker wants a pair of sequential pictures–an ultrashort movie, essentially–that will mark the progress of the light over a short period of time.
The laser light will travel through a series of spaces as it bounces through angled mirrors that face each other in pairs along a 9-foot-long table. A 7-foot-long pulse of light will travel through the mirror course in 90 nanoseconds.
With a camera capable of shooting 10-nanosecond frames, the trick is figuring out which 10 nanoseconds to capture, there being 90 million such periods in one second alone. There is little time for reaction lag, and even though the laser’s hardware generates a pulse to trigger Nebeker’s camera at the same time that it triggers the laser, the circuitry and the 20-foot cable attaching laser to camera add nanoseconds of delay here and there.
In order to capture the image, Nebeker has to measure and synchronize the delay of the laser with the delay of the camera. Employing a trial-and-error process, he ultimately gets the camera to switch on precisely while the pulse is passing through the mirrors. As predicted, a 100-nanosecond exposure produces an image showing, in the first frame, the entire 90-foot course lit up; the second frame is dark. He then cuts exposure time down to 50 nanoseconds, which produces a first-frame image showing half the course illuminated. The second frame, however, remains dark. Tweaking the delay between frames, he produces a second frame that shows the tail end of the pulse disappearing into the terminating “beam dump” box at the far end of the mirror course.
Nebeker adjusts the exposure time to 10 nanoseconds, the camera’s limit, and the attending Spectra-Physics engineers begin to get excited about what they’re seeing. As expected, the computer screen shows a first frame in which only one leg of the mirror course has laser light in it. The rest is dark. In the second frame, also 10 nanoseconds long and exposed after 10 nanoseconds’ delay, a leg of the course two over from the first is illuminated. Nebeker has captured two discrete portraits of a laser light beam traveling within the 90-foot course.
The grainy black-and-white digital images would not pass muster with John Woo. Nonetheless, they show light the way science fiction has long depicted it–moving as a discrete packet across a short space (think of Kirk’s phaser, set to stun, in the old Star Trek). The small crowd of engineers is delighted. “I think we need one of those!” one engineer calls out as he looks longingly at the Cordin Model 220 camera.
This was a gee-whiz shot, no question: The Cordin camera managed to freeze light in its tracks, as if it were no faster, say, than a speeding bullet.