Today, high-definition, multiformat, multi-frame-rate cameras are the norm. Photo courtesy Panasonic.
The modern digital television camera is an impressive piece of technology. In fact, it might be the ultimate analog-to-digital converter. It converts light-intensity variations into an analog electrical signal, and then converts that to a digital signal. Given the high bandwidth of the video signal and the fact that the digital conversion is three dimensional (one temporal dimension and two spatial dimensions), the digital video camera ranks high on the scale of commercially available converters. And the range of resolutions available in modern video cameras is huge, from Webcams to eight-megapixel electronic cinematography cameras and everything in between. Some scientific image sensors run at high frame rates; some hang on the back of telescopes and are cooled by liquid nitrogen to reduce noise and increase sensitivity. Of course, such devices don't serve broadcast television. But the research-and-development funding for such advanced sensor technology has over time trickled down to mainstream television applications, resulting in improved sensitivity and resolution, effective methods for managing defective pixels, and other important developments.
High-resolution image sensors like this 12-megapixel CCD chip represent the upper limit in electronic imaging technology. Photo courtesy Canada France Hawaii Telescope.
Imaging technologies
Image sensors designed for scientific and technical applications define the upper limits of electronic imaging technology. It is important to recognize that, though HDTV is just now beginning to impact consumer products, HDTV cameras are well into their third and fourth generations. At the same time, some research institutions have begun to define extremely high-definition television for the next generation of camera technology. A 12-megapixel image sensor might be the seed for research into real-time imaging at stunning quality levels. But remember that HDTV's native resolution is already adequate for most of today's consumer applications.
Image sensor technology has advanced tremendously since the days of tube cameras. Plumbicon, orthicon and other tube technologies served well in their day. But factors such as lifetime operating cost (capital and replacement cost), sensitivity, labor cost for constant adjustment, size, power consumption, weight, limiting resolution, and modulation transfer function have consigned tube cameras to the scrap heap of TV history. Early CCDs struggled to match the performance of modest plumbicon cameras, but CCD technology quickly advanced to eclipse tube technology and has become today's standard. The newest technology for image sensing is CMOS. These image sensors offer lower noise, higher sensitivity and other improvements. CMOS image sensors are showing up in consumer still cameras and in both consumer and professional video cameras. It is logical to expect that the research dollars invested in developing CMOS image sensors for consumer applications will drive it to higher levels of performance, challenging CCD's dominance in professional cameras.
A glimpse inside this old RCA TK-60 camera reveals a 4-1/2-inch orthicon tube image sensor. Photo courtesy Pavek Museum of Broadcasting, MN.
The ultimate criterion for imaging devices is the distance between display pixels. The human visual system can distinguish adjacent pixels only when the distance between them is at least one arc minute, depending on age and several other factors. One current HDTV format, 1920×1080 pixels, satisfies the limits of the human visual system's resolving power when viewed from approximately 3.3 times the picture height. (For NTSC, the corresponding viewing distance is about six times the picture height.) An image created with twice as much horizontal and vertical resolution, using square pixels, would either increase the allowable viewing distance or increase the picture's surface area by roughly four times. Such displays would not be very useful in most consumer environments, but could be attractive for virtual reality and other applications. They are aimed at futuristic markets or industrial applications. But the purpose of this article is to review real-time imaging systems.
The lens
It is also important to consider a television camera and its lens as a combined system. The camera and lens contribute equally to picture quality. For instance, buying an HDTV camera and using an older lens with a modulation transfer function that is not a good match to the performance of the camera's imager will yield inferior results, or at the least impaired performance. Larry Thorpe, former Sony executive and advocate for HDTV for nearly two decades, was recently asked — only slightly facetiously — if his current position with a lens manufacturer might lead him to view television cameras merely as a lens accessory. Indeed, some manufacturers have reduced the size of their high-quality television cameras until they are significantly smaller than the lens to which they are coupled. Modern HDTV system performance would not be possible without the incredible advances in lens performance over the last decade. But such performance does not come cheaply; long-zoom-ratio HDTV lenses still cost between $150,000 and $200,000.
The lens is just as important as the camera. Photo courtesy Thomson Grass Valley.
The combined camera-lens system affects many aspects of performance, including colorimetry. The combination of image-sensor sensitivity and the spectral bandpass of the optical system is additive. In general, lens manufacturers design their products to provide a flat spectral response to be compatible with any other manufacturer's cameras without modification. The only customization on the lens is its electronic interface to the camera. But, today, SD/HD-switchable cameras increasingly dominate the marketplace. So, when buying a lens, you must consider carefully whether the lens should have a ratio converter that will allow the taking angle of the lens/camera system to remain fixed in both 16:9 and 4:3 aspect-ratio imaging modes. If the lens is intended for use with a specific image sensor at 4:3 and the camera is switched into 16:9 mode, two effects can happen. The image becomes significantly narrower for a given focal length. And, if poorly adapted to the application, vignetting can occur, especially at wider focal lengths.
For modern three-chip CCD or CMOS cameras, the mechanical alignment of the three color channels can be close to perfect. With tube cameras — even early HDTV cameras — chromatic aberration was not nearly as critical because it often was masked to a certain extent by the performance of the imaging tubes. To be equal partners in today's improved camera systems, lenses have had to improve along with the image sensors. Manufacturers of today's lenses have dramatically reduced the effects of design compromises that were significant in lens technology only a few years ago. In addition to reducing chromatic aberration, they also have reduced focus breathing. Focus breathing appears as zooming when users adjust the lens' front focus. All lens manufacturers have improved the performance of their products in large part to satisfy the changed characteristics of today's cameras.
Canon and Fujinon both have presented papers at technical conferences in the last several years to discuss improvements in lens technology as they apply to both television and electronic cinematography products (TV zoom and cinema prime lenses).
Sony’s HDC-910 can capture 1080i 50/60 images and output them as 1080i 50/60, 480i 60 or 720p 60.
One recent paper described experimental technology to allow automated focus for HDTV camera systems. It uses advanced electronics in the lens system, including a second imager. The purpose of the imager in the lens is solely to sample the image and determine when the captured image is in sharpest focus, and then provide feedback to the focus servo to correct errors. This may become critically important in some future applications. HDTV applications suffer from an apparent reduced depth of focus, making it more difficult for camera operators to maintain optimal focus, especially when the lens is operating at wide aperture settings as in a night sporting event. An automated focus assist might provide a more pleasing result for the viewer.
One important distinction between HD and SD cameras is the significant difference in cost and performance. Some manufacturers have begun to whittle away at the problem by designing image sensors that are inherently oversampled for all resolutions. Thomson Grass Valley employs this technique and has sold many HDTV cameras that are capable of native resolutions from SD through 2.4:1 aspect ratio HDTV for electronic cinema applications. This technique uses vertical oversampling while maintaining 1920 horizontal pixels. It permits a native 1080 horizontal resolution. And, by combining samples before processing the picture, it permits other horizontal resolutions, such as 1280 samples for 720p and 720 samples for 525/625.
The same concept can apply in the vertical dimension, combining multiple vertical samples to achieve 1080-, 720-, 525- or 625-line outputs. This technique of oversampling in the vertical dimension is both clever and effective at achieving native resolution. A purist might argue that it is not truly native resolution because the samples are not sited in the precise locations that a native sample might be, but it's still a good approximation and yields good results. One argument put forward by other manufacturers is that such a technique comes with a sensitivity penalty. Other manufacturers are known to be working toward similar variable-aspect-ratio and variable-resolution image sensors and camera systems, so look for even more options.
Oversampling
Other factors also differentiate the wide variety of television cameras. The biggest factor is price. The differences between many current production cameras and high-end consumer cameras include marketing, the amount of metal in the case and the quality of the lens. High-end consumer cameras that use three image sensors can produce high-quality pictures indeed. Many producers of entertainment and documentary programs have put such lower-cost hardware in the hands of capable videographers and achieved stunning results. The cost secret is the hidden fact that the development of the consumer hardware is amortized across perhaps millions of delivered cameras, while the professional camera is sold at higher margins but delivered in quantities of hundreds or perhaps a few thousand. Usually, the biggest factor separating these cameras is the optics. A $4000 consumer camcorder cannot have a high-quality lens without raising the cost. But, put a $4000 lens on a high-end consumer camera and it will be clear (no pun intended) that the results are spectacular for the price.
Occasionally, feature creep runs in the other direction. Initially, the only 24fps cameras on the market were high-end HDTV cameras. Then, Panasonic introduced a professional SD model for under $5000 that could record 24p images. Now, the same feature has shown up in consumer hardware, at an up-cost to the consumer, but little additional manufacturing cost. This kind of product differentiation is not unique in the broadcast business. Indeed, helical-scan VTRs existed in the professional domain before the consumer electronics industry in Japan brought us Betamax and VHS. Today, the same crossover of features and technologies is making products much more capable. For example, DVD camcorders arrived on the market soon after the cost of DVD burners for computer applications came down to a modest amount. Volume drives the manufacturing cost, and features sell the new application of the hardware.
Cost versus performance
Two decades ago, the difference in image quality between a studio camera and a handheld was substantial. Handheld cameras were versions of their big-brother studio cameras. To cut size and weight, camera makers sought a practical package by eliminating everything unessential. It is not unreasonable to compare the picture quality of some early (and expensive) handheld professional cameras to the image from today's three-chip consumer camcorders. In fact, the current consumer products probably produce a superior picture. Interestingly, the difference in price between these two types of cameras may be about two to four orders of magnitude.
Today, the difference in image quality between portable and studio cameras is less dramatic. The electronics are simply so small that housings larger than that of a portable camera are not necessary. Digital processing, solid-state image sensors and digital transmission make a large package unnecessary. Some manufacturers have taken to building only portable cameras, notably Thomson Grass Valley, and resorting to “sleds” to mount the camera body to larger lenses. Here, the camera truly becomes a lens accessory, with the rear of the lens often holding the camera weight and the sled simply holding the lens. By doing this, manufacturers have minimized the cost differential between applications, and companies needing both types of cameras can buy a quantity of sleds to match their collection of long lenses without having to duplicate expensive electronics in camera heads that might not be used for every show. Because the number of cameras that broadcasters use per production has increased steadily over the years, any flexibility in camera cost makes the mobile television business more affordable.
Studio versus handheld
This approach also makes available the full feature set of a studio camera in all applications. Return video loops, intercom, tally, prompter outputs and other features are no longer excluded from applications that require a small (but not handheld) package, which increases production flexibility greatly. Lastly, this approach is clearly the best solution for technicians because now they can focus on the technology of fewer products. It also reduces repair time and parts inventory. Counterbalancing this is the long-held opinion of some camera operators that the low mass of buildups makes high-quality camera moves more difficult. Perhaps this is true. You might remember when RCA sold a TK-76 portable camera and a TK-760, which was nothing more than a TK-76 with a big case around it to make it feel big. That might not satisfy the market today, but the issue is one you must consider when specifying studio cameras.
Finally, with zoom ratios extending beyond 100x, the stability of the camera-lens system is a serious consideration. When the camera's mounting platform cannot prevent vibration — for example, on a scaffold at a stadium — the mass and inertia of a large camera can help produce a more stable picture. Even with modern lenses that have internal image stabilization, this total body mass may be an important supplement in some applications.
Editor's note: For a full discussion of imaging and critical human visual system capabilities, see the April 2004 article, “HDTV displays: How good do they need to be?” by Jukka Hamalaien, available atwww.broadcastengineering.com.
John Luff is senior vice president of business development AZCAR.
