Angles Of View

The television and video worlds have been talking about HDTV and its 9:16 aspect ratio for so many years now that the rest of us may be forgiven a little skepticism and a lot of impatience. Yet, finally, it appears that HDTV really is on the horizon and 9:16 displays really are about to become widely known. But, what is it about 9:16 that makes it so special? Why, for instance, aren't the screens for the future going to some other aspect ratio, say 9:17 or 8:16? The truth is that a great deal of very careful work went into figuring out this aspect ratio and its properties. It may well, therefore, be useful to continue

Looking at 9:16 - In the Background

Although the concept of HDTV first acquired currency more than 30 years ago, the path followed by the development of its technology has been anything but straight. Nevertheless, although quarrels over precise definitions and standards abounded in both the national and international communities, the overarching conceptual goal was always clear. Television was to be made over so that it could look like the movies.

Three decades later and at the end of the millennium, display and presentation technologies parallel to television have become so powerful and sophisticated that there is little question that realized HDTV will have applications far wider than solely the broadcast industry. Obviously, the most prominent of these collateral businesses is the computer business which not only has lots of ideas of its own about High Definition but almost certainly has designs to subsume television itself.

Having said that, it is worth noting that the original thinking that went into both conceiving and defining HDTV paid no attention whatsoever to such extraneous and, at the time, superfluous concepts as data display. Instead a group in Japan called Nippon Hoso Kyokai (NHK) set out in the late 1960's to figure out just what it would take to deliver an entirely "new viewing experience" to television watchers. In pursuit of that goal, NHK arranged that a large number of technically untrained viewers were shown a broad variety of electronic images whose parameters were made to differ greatly, one from another.

Much of this work was repeated in the United States and Canada during the mid-90's. Even if the methodologies were disparate, the conclusions of both data sets were convincingly similar.

In the United States, for instance, a lot of research was undertaken to establish that Americans tend to watch their television sets from a distance of about ten feet. Nowadays, with larger, direct view CRTs readily available, that number may be reexpressed as equivalent to 7 screen heights, but the idea is unchanged. The point to see here is that at such a distance most of the limitations and artifacts implicit in the standard NTSC signal can't easily be noticed.

On the other hand, the "window" through which we look at a televised world is only about 10º wide. Lurking within this constriction is one of its primary determinants: resolution. With only 525 lines to work with, NTSC (625 for PAL and SCAM) doesn't do a very good job of furnishing much detail in shots which we would describe as being wide-angle. It needs instead to rely on its ubiquitous close-ups.

Thus, when it comes time to establish what a better, less limited picture would be, we can recognize that the new format's size and shape will be inextricably dependent on its resolution and vice versa. Let's unbundle the two for a moment and see how the HDTV standard emerges from both.

From the point of view of viewing angle, the worst thing about 10º is that it fails almost entirely to include peripheral vision. In contrast, even a modestly sized movie screen subtends closer to 30º of our visual field and is thus very much more a compelling and involving visual experience. When we come to transfer that angular relationship out of the cinema and into the living room, we discover that, at a ten-foot viewing distance, the screen has grown in height from something like 17 inches to more than 36 inches.

When we take those 36 inches and multiply by the HDTV aspect ratio of 1.777, we get a width of just more than 64 inches. And, to bring the analysis full circle, an image 64 inches wide at a viewing distance of 10 feet subtends almost exactly 30º. What is consistent between the two televisions is importantly neither their size nor their shape; it instead is their expected viewing distance — the same living room friendly 10 feet.

Of course, the new screen is a whole lot bigger (to say nothing of wider) than the one in our old TV and so our ten feet is no longer 7 picture heights back, it's a mere 3.3 picture heights. And, at only 3.3 heights back, we're going to have no difficulty at all detecting the evident coarseness of the old, NTSC resolution.

Quite a number of previous articles in this series have discussed the resolution of visual displays extensively. Except for occasional cavils, the position advanced in those pieces was that more resolution is generally preferable to less. The developers of HDTV, however, approached the resolution question somewhat differently. Once again, they started with viewing distance and concluded, perfectly reasonably, that the sensible limit for the resolution of their display was at the point where the eye is able just to make out the finest details in its picture. More resolution than that, they saw, would just be profligate.

The optimal viewing ratio, then, is formed by relating viewing distance to picture height. Why height and not width? Well, it turns out that "the discernable detail [of a television picture] is limited by the number of scanning lines presented to the eye and by the ability of these lines to present details separately."¹ Obviously, not all the smallest details (by which we mean one pixel's worth) in a TV picture are going to fall within a single scan line. Some are going to overlap into two scan lines and when this happens, of course, some vertical resolution is inevitably lost.

To see why this effect is important, we need to review briefly that there are two distinct ways in which those scan lines are written, interlaced and progressive. If the vertical resolution of a system is, say, 1080p (for progressive) it means that the raster starts at the top of the screen and, like a typewriter typing single spaced, scans across line 1, then 2, 3, 4 … all the way down to line 1080 before zipping back to the top of the screen to start again. If, however, the vertical resolution is given as 1080i (for interlaced), it means that the raster starts at the top of the screen and, like a typewriter typing double spaced, scans across lines 1, 3, 5 … down to line 1079 before zipping back up to the top of the screen to start typing the even numbered lines 2, 4, 6 … 1080.

For reasons having to do with bandwidth conservation, conventional (NTSC) television is interlaced. For reasons having to do with maximizing resolution, computer displays are generally progressive. Although both scanning schemes result in some loss of vertical resolution (typically about 30%), interlacing is worse (about 50%).² It had been an early goal of the HDTV developers to do away with interlacing, but, at some scan rates, it remains, for technical reasons, within the finalized standards.

After studying lots and lots of alternatives (both interlaced and progressive), the Japanese NHK group concluded "the preferred distance for viewing … their system has a median value of 3.3 times the picture height, equivalent to a vertical viewing angle of 17º."³ For that ratio, it was found that just more than 1,000 scan lines were optimal.

The rationalization of the width of the HDTV aspect ratio emerged from somewhat different considerations. The movie industry had more or less invented the widescreen concept as a way of fending off what it initially viewed as television's threatening encroachment of its central position in the entertainment industry. Film, of course, is an analog medium and, therefore, has symmetric vertical and horizontal resolution. When you stretch it horizontally, however, its resolution in that dimension is proportionately diminished. The widest of popular film formats, CinemaScope, reached the limit of this stretch and was 2.35 times wider than its height.

Again, for reasons of bandwidth considerations, the developers of HDTV chose not to go that far and settled instead on a width that was 1.777 times the height. This translated to a 30º horizontal viewing angle at a viewing distance of 3.3 picture heights. Although 30º is only about 20% of our horizontal visual field, it happens conveniently to cover the area of the field "within which most visual information is conveyed."⁴ Thus, the 9:16 format came to be.

Throughout the genesis of HDTV and 9:16, one thing has remained admirably clear: this technology has been designed around the programming and not, as has so often been the case in the past, the other way around.⁵ As we prepare to welcome this new format into our A/V world, however, is not the exact reverse the case?

¹ Whitaker, Jerry, DTV: The Revolution in Electronic Imaging, McGraw-Hill, 1998. p. 71.
² ibid.
³ ibid. p. 72
⁴ ibid. p. 73
⁵ ibid. p. 43