poynton@vector.Sun.COM (Charles A. Poynton) (03/29/89)
Reprinted with permission: >From stefik.pa@Xerox.COM Tue Mar 28 06:45:58 1989 >Subject: Parameterized digital standard >To: poynton@sun.com > >As an observer with a casual interest in HDTV, I've been puzzled by the >attention and worry given to defining a standard in terms of fixed >numbers. > >Pardon my ignorance, but why not have a digital standard for HDTV and why >not make it parameterized. So that the signal itself contains data about >refresh rates, pixel size, horizontal and vertical lines, number of bits >of color, and anything else you want? > >Receivers would then need to have the logic to cope with what they got and >render it as best they can -- clipping, transforming, averaging, or >whatever. > >For one thing, it would be nice to be able to have very large screens >which also have very high resolution. It may be practical to have >specialized displays, of different prices, suitable for rendering >different kinds of images. > >This attitude about digital processing strikes me as something that >computer manufacturers bring to the table which may be outside the >thinking of the broadcast people, who seem to be in the driver's seat. > >----- mjs In the computer graphics area, parameterizing software such as renderers makes a great deal of sense, and is fairly easy to accomplish. Also, defining interface standards to convey information about representation is very wise. Xerox knows a lot about this from their document interchange work. TIFF 5.0 does a remarkably good job at providing for exchange of image parameters, even to the extent of conveying colourimetry data (which is something that even "high end" computer graphics systems rarely provide). However, when it comes to constructing hardware, parameterization is much more difficult. Take the example of multi-scan monitors. A CRT monitor intrinsically is incapable of variation in horizontal scan rate of any more than perhaps +-5%. This is because the monitor's high voltage, which accelerates the electron beam toward the screen, is derived from the horizontal scan. Many, many parameters of the monitor, such as focus and convergence, depend on this high-voltage system. It is technically possible to build a monitor which operates over a wider range of scan rates than +-5%, as is the case with the multi-scan and auto-scan monitors now widespread in the low-end PC graphics world. Some of these accommodate a factor of two of horizontal scan frequencies. However, this "parameterization" comes at a price: some number of deflection system components are switched in and out of the circuit by relays. As Louis Chevrolet is reported to have said when he invented the clutch, "It's brutal, but it works." This circuitry adds more than a hundred dollars to the price paid by the end user for the monitor. But the feature delivers essentially no utility: it exists primarily due to the absence of standards in the display interface area. What the user really wants is to buy the best monitor that he can afford, plug it in to the best framebuffer (graphics board) that he can afford, and have the two work together at the highest possible resolution all the time. Also, there are certain parameters that are either extremely difficult or impossible to alter for some devices. For example, a CRT monitor may be adaptable to various line counts, scan rates, and aspect ratios, but CCD cameras, plasma displays, and LCD displays have their pixel count and shape physically etched into the device and therefore fixed for all time. Finally, the economies of high volume production weigh in favour of tightly specifying parameters for certain devices. Dumont Electronics in the late 1940s proposed that the FCC should adopt a "standard" for broadcast television which could vary over a wide range of line and frame rates with a given video bandwidth. It is plainly clear today that television would not have been able to flourish in that environment because of the added expense and system complexity. Motorola tried "works in a drawer" as a product concept for Quasar television sets, but it turned out that the necessary connectors added cost to the receiver which the consumer was unwilling to bear. MIT are currently proposing an "open architecture receiver" for HDTV, but the practicality of the proposal has yet to be demonstrated. I believe that it would be a great advantage to the computer industry to be able to utilize the same monitors that will be manufactured for HDTV. Ensuring that the 1125/60 standard is applicable to the computer industry is a necessary first step for this to occur. C. ----- Charles A. Poynton Sun Microsystems Inc. <poynton@sun.com> 2550 Garcia Avenue, MS 8-04 415-336-7846 Mountain View, CA 94043 -----