|
|
 |
|
|
 |
|
 |
 |
|
Moved But Not Shaken
|
|
by Mark Schubin
|
This year is divisible by four, which means it features both U.S. Presidential election events and Olympic Games. Candidates are said to run for office, and athletes run, jump, dive, or move in other ways. Some of this fall's competitors in the Beijing Paralympic Games might be blind, but other athletes -- and all of this year's Presidential candidates -- seem, despite all of that running, to do fine with a literal and physiological interpretation of what George H. W. Bush, at the onset of his campaign for our nation's highest office, called "the vision thing." Too bad video cameras don't function as well.
The first Olympic Games to be televised were those in Berlin in 1936. Three electronic cameras were used. They were called Fernsehkanonen, literally "television cannons," which should give some idea of their size and weight. The cameras didn't move; the athletes ran past them.
Jesse Owens, the American who won four gold medals at those games, might have seen those cameras. Certainly, he saw the lane of the track within which he ran. But how?
If you attach a video camera to the head of a runner, the result will probably be an image so shaky as to be unviewable. Humans can see clearly under such conditions only because we have superb image stabilizers built into our brains.
Try these two experiments. First, close one eye and look around. Your eye is moving, but nothing that you're looking at appears to move. Now take your finger and press gently on the lid of the open eye, enough to move the eyeball. Suddenly, what you're looking at does move.
For the second experiment, sit in front of a picture tube showing a video image. Then bite into a very hard pretzel or carrot. Chances are excellent that you'll see what appears to be a "hit" or "glitch" in the picture.
What's happening is that your brain receives information from many parts of your body to indicate what the position and motion of the eyes is, and then it processes their images to remove that motion. The processing works so well that you can even read a sign while jumping on a trampoline -- but only after the first few bounces or so, while your motion-analysis system figures out how the trampoline works.
That's why pushing on the eyeball or biting on a carrot can cause unstable imagery. The brain isn't accustomed to an eyeball push and doesn't know the compression strength of a carrot. Worse, as far as videographers are concerned, it doesn't know what motion caused an already-acquired picture to be shaky. That's why the built-in motion-stabilization system that allows us to read while jumping won't correct the video of a camera attached to us when doing the same jumping. So other image-stabilization techniques are required.
The television cannons of the 1936 Berlin Olympics used one such technique, anchoring a camera to a fixed and stable mount. Today's tripods are smaller and lighter, but they accomplish much the same function. Dollies, pedestals, jibs, and cranes offer some camera motion with stable pictures, as long as the floor, track, and/or moving joints are smooth and even. But there can be a stability problem with even a fixed, anchored mount.
Consider NASA launches. The Space Shuttle is quite large, but it also moves very fast. Keeping it as more than just a bright dot in the picture requires a lens with a very narrow acceptance angle, a small fraction of one degree. Visually imperceptible jitter of the camera mount could mean the difference between the spacecraft being in the picture or not. Yet NASA captures gorgeous, stable HDTV images of the Space Shuttle, even shot from a moving aircraft from dozens of miles away. That seeming miracle is accomplished by stabilizers.
There are three basic types of stabilizers. The oldest works on the camera mount itself.
In some cases, the goal is simply to isolate the camera from an unstable platform. Early vehicle mounts used springs and shock absorbers to smooth a camera's ride in the same way that springs and shock absorbers smooth a passenger's ride in a car.
That's often sufficient for many lens angles. Steadicam mounts, for example, which revolutionized both cinematography and videography, isolate a camera from its operator's body, allowing it effectively to float. Even some helicopter mounts simply isolate the camera from the vibrating aircraft.
Such isolation is fine when the lens acceptance angle is large. At the wide end of a popular wide-angle lens used for videography (4.5-mm focal length in a 2/3-inch camera format), the acceptance angle for widescreen (16:9) images is roughly 94 degrees horizontally by 62 degrees vertically. A jitter that swings the camera roughly one degree horizontally would displace the image by only about one percent.
When tracking a 122-foot Space Shuttle at a distance of 25 miles (132,000 feet), however, the acceptance angle is roughly 0.053 degrees. A shift of less than a tenth of one degree means the difference between seeing the spacecraft at all or not. So NASA needs better stabilization than springs and shock absorbers.
For many years, it seemed videographers did not. Moving-camera systems -- everything from shoulder-mount to Steadicam to cranes and jibs -- tended to be used with wide-angle lenses. Not only did that make vibration less noticeable, but it also added to the visual impact of the moving camera. Objects in the foreground were larger and moved more than those in the background, adding depth cues to the images.
Lenses with longer focal lengths tended to be used in sports on fixed, stable mounts. But, as the acceptance angles of those lenses got narrower and narrower, even the best videographers had difficulty keeping jitter out of the pictures. One popular sports field lens has a maximum focal length of more than 2300 mm in the 2/3-inch format. For widescreen shooting, that's an acceptance angle of just about 0.24 x 0.13 degrees. Never mind the Space Shuttle 25 miles downrange; how does one keep a football in position at such a narrow angle? And then there are those quadrennial events, the elections and the Olympics.
For a press conference, a news videographer can use a tripod or even just plug into a pool-feed camera signal. Following a candidate through a crowd, however, probably means shoulder-mounted coverage, and, with hundreds of others in the way, it might require a tight lens angle.
In the Olympics, there's a different kind of pool feed. A television director might want to zoom in tight on the face of a swimmer from a camera on a remote-controlled moving dolly tracking the length of the pool.
NASA uses gyroscopes to assist with their mount stabilities. The spinning components fight vibration. They work not only on the fixed mounts used on NASA's land-based tracking cameras but also on aircraft-based mounts.
Axsys Technologies' Cineflex V14HD is an example of one of the latest versions of such gyro-stabilized camera systems. It has five-axis stabilization and can deal with 2/3-inch-format HDTV focal lengths ranging from 4.5 to 3520 mm. It can spin a full 360 degrees, has an elevation range of 210 degrees, and can deal with intentional pan and tilt changes of up to 55 degrees per second, with a remote-controlled positional accuracy of a thousandth of a degree.
There are only three things about it that might concern a typical videographer: It's not cheap. It weighs 68 lbs. And it consumes 170 watts (with peaks of 230 watts).
None of that should concern an Olympics broadcaster. An independent documentary videographer following a candidate, however, would have a hard time shouldering 68 lbs., not to mention the weight of the batteries necessary to provide 170 watts continuously for a long period. And then there's the Metropolitan Opera.
Starting with The Magic Flute late in 2006, the Met has been using moving camera systems for its Live in HD transmissions to movie theaters worldwide. Each "cinemacast," however, is also a performance in front of an audience of thousands in the opera house, some of whom paid hundreds of dollars for their tickets. Those non-video-viewers don't want to be distracted by the equivalent of a helicopter mount zipping back and forth across the bottom of the stage.
For those applications where mount stabilization doesn't seem to work, there's image stabilization. As early as 1915, a patent application was filed covering an optical component inside a telescope that used its own inertia to counter image vibration in a manner similar to the way Steadicam counters camera vibration.
Dynasciences brought that concept to moving-image shooters as the Vibra Stop lens. Just as gyro-stabilized mounts are more appropriate than Steadicam for narrow-angle shooting, however, Dynasciences' Dynalens could also work better than the Vibra Stop for longer focal lengths.
The Dynalens, first sold in 1965, was a fluid-filled bellows between two sheets of glass. Gyro-controlled motors could squeeze the bellows to form a prism to bend light to counter vibration. Some of the latest optical image stabilizers are Canon's IS-20B II, which mounts to the front of a lens, and Fujinon's TS-P58A, which mounts to the rear. The Met has used both on the moving-camera systems of its global cinemacasts.
Sometimes, however, even lightweight, optical image stabilizers are too heavy, too power consumptive, too expensive, or simply not available, and the same may be true of stabilized camera mounts. In that case, there's always the third category of image stabilization, the digital version. Unlike the other two kinds, it can remove shakiness from even already-shot video.
Although it doesn't date back quite as far as stabilized mounts or optical vibration compensation, digital image stabilization isn't exactly new, either. At the 1988 convention of the National Association of Broadcasters, Toshiba demonstrated a system developed jointly with NHK, the Japan Broadcasting Corporation.
Smoothcam is a software version that's part of Apple's Shake extension to Final Cut Studio. For-A's IVS-700HS is a hardware-based system that can remove jitter as extensive as 40% of the picture size and be used live. It automatically decides what is shaky and leaves the rest alone.
Of course, removing jitter after the fact is possible only if a shutter was used so that individual frames are not blurry. Or is it?
Before digital image stabilization, before Canon's or Fujinon's optical stabilizers, and before Steadicam, an article published in the Proceedings of the IEEE (Institute of Electrical and Electronics Engineers) in July 1972 suggested that deconvolution could be used to sharpen images that are blurry because of unstable capture. An example of a photo of a highway sign shot handheld from a moving car was brought from unreadable to readable, if not to the look of a photo shot from a tripod-mounted camera on a concrete platform.
It has been more than 35 years since that paper was published. Deconvolution has since been used (by Thomas Stockham, author of the 1972 article) to remove the effects of the amplification horn from mechanical recordings of the great tenor Enrico Caruso, so listeners can hear what he might have sounded like at the Metropolitan Opera House.
Signal-processing technology has greatly improved in the intervening years. Might it now be possible to ignore camera and image stability altogether? Might a videographer shoot haphazardly and then run the results through a digital processor to come out with professional results?
In a word: no.
advertisement
Digital Display Rebates From SYNNEX
Resellers Can Bundle and Save On Digital Signage Solutions From SYNNEX -- HP, NEC, LG & Samsung. Audience Voting Equipment for Interactive Meetings & Events
Create exciting audience polling & game show events using wireless group response keypads and other audio visual presentation equipment. Buy or rent. Audio and Video Cables
Features HDMI, DVI, VGA, SVGA, KVM, EVC, and S-Video cables and adapters. Tons of free information on the site along with tips, tutorials, and more.
|
|
|
|
Videography is part of the NewBay Media Community
copyright 2007© NewBay Media, LLC. 810 Seventh Avenue, 27th Floor New York, NY 10019 T (212)378-0400
F (212)378-0470
|
|
|
