Why Do We Interlace?

We're familiar with interlaced scanning; we know that it has its problems, but we have to live with them in the NTSC world. What about DTV? Do we have to live with the problems of interlace in the DTV world?
Author:
Updated:
Original:

One of the benefits that we derive from DTV is the choice of scanning formats and frame rates it affords. For some time, much television production and post production has been done in the component domain. It is, however, only possible to transmit one video format in analog television in North America: NTSC.

We know the NTSC parameters in the U.S.: 525 total lines per frame, about 480 active lines per frame, each frame composed of two half-frames or fields, a vertical repetition rate of about 29.97 frames or 59.94 fields per second. DTV opens the door to many more scanning formats, and it is inherently a component-based, rather than a composite-based system. DTV scanning formats may be categorized as HDTV and SDTV. In HDTV, two scanning formats are readily available to the broadcaster, 720p and 1080i, while in SDTV, 480i and 480p are available (some call 480p extended definition or EDTV rather than SDTV).

Each of these formats may be used at several frame rates. The interlaced formats, 480i and 1080i, are transmitted at 60 or 59.94 fields per second, which is 30 or 29.97 frames per second. The progressive formats may be transmitted at 30/29.97 frames per second, 60/59.94 frames per second, or 24/23.98 frames per second, the frame rate of most film used for television and theatrical projection.

Because most DTV broadcast facilities operate within large plants based on NTSC television, the 1/1.001 timing factor is used universally in U.S. DTV broadcasting. Thus, when we speak of 60 frames per second or 24 frames per second, we really mean 59.97 or 23.98 frames per second.

INTERLACE IN DTV

We're familiar with interlaced scanning; we know that it has its problems, but we have to live with them in the NTSC world. What about DTV? Do we have to live with the problems of interlace in the DTV world? It is logical to ask why we use interlaced scanning at all. Where did it come from, and what gave rise to it?

In the 1930s, television was in an experimental stage of development. The earlier experimental electronic television systems used progressive scanning. This seems logical, as progressive scanning is just scanning the lines of the television picture one below another; Line 1 followed by Line 2, followed by Line 3.

The developers reached a point at which they decided that in order to produce a picture with adequate resolution, at least 400 lines were required in a frame. To meet the desired line count, the progressive frame rate had to be restricted to about 30 frames per second in order to stay within the 6 MHz bandwidth of a television channel and leave room for a sound signal. The problem was -- 30 frames per second -- or 30 large-area light flashes per second is insufficient to prevent large-area flicker, which is the sensation of the picture perceptibly fluttering or flashing. Flickering occurs when the vertical repetition rate (the number of light flashes per second) is too low for the specific viewing circumstances.

The critical flicker frequency, or the repetition rate above which flicker is not perceived, falls between 40 and 60 repetitions per second. The exact flicker threshold depends on factors that include picture brightness and ambient lighting, but 30 flashes per second is below it under any viewing circumstances. Although theatrical motion pictures run at a rate of 24 frames per second, each frame is projected twice, raising the flash rate to 48 per second. This is above the critical flicker threshold for relatively low-brightness images in a dark movie theater, but it is well below it for bright television pictures viewed in lighted rooms, as may be attested by anyone who has viewed PAL television with its 50 flashes per second.

Speaking more precisely, 48 flashes per second is above the flicker threshold in a dark movie theater for foveal, or straight-ahead viewing, but not for peripheral viewing. This is because foveal viewing by the human eye is largely done by the cones. Most of the cones with which we perceive color and detail are located just behind the lens of the eye, where they receive the light that comes into the eye from a straight-ahead direction. The rest of the retina is covered with rods, which receive most of the light that does not enter the eye from straight ahead, and are dominant in nonfoveal vision.

Rod vision is coarse but acute; the rods do not perceive color or fine detail, but are much more sensitive to movement than are the cones. This fact led to several of the big Todd-AO productions from the golden age of movie making being shot with parallel 24 fps and 30 fps cameras. The wraparound Todd-AO screens involved the viewer's peripheral vision so much that there was excessive perceptible flicker at 48 flashes per second. For Todd-AO projection, the 30 fps film was used to raise the flash rate to 60 fps, while the 24 fps film was used for standard projection to flat screens. By the way, dogs' eyes have no cones, only rods, so although Fido can detect the tiniest movement, he would probably not care for movies unless he could see them in Todd-AO.

INTERLACE TO THE RESCUE

The flicker problem in television was solved by using an innovation called interlaced scanning. In interlaced scanning, each picture or frame is scanned as two half-frames or fields, each field containing every other line of the frame. In Field 1, all the odd lines of the frame, 1,3,5, etc., are scanned, while in Field 2, all the even lines, 2,4,6, etc. are scanned. The lines of Field 2 fill in the blanks left when Field 1 was scanned, and vice versa. The fields are scanned, transmitted and displayed sequentially. As they are sequentially displayed, the human vision system perceives the odd-numbered lines and the even-numbered lines to be interwoven or interlaced, which integrates them into a complete picture.

This seems, on the surface, to be the best of all possible worlds. Thirty frames' worth of picture information is transmitted each second, while the vertical repetition rate is doubled to 60 light flashes per second. Being people of the world, however, we know that there is no free lunch. The goal of interlaced NTSC was to effectively provide about 480 lines of vertical resolution, while keeping the vertical repetition rate above the critical flicker threshold. It was rather quickly determined that while the latter goal was met, the former was not. This is true because the full resolution of an interlaced picture is only realized when it is a still picture. The still picture is the best case; when the picture moves vertically between fields, vertical resolution is compromised. In the worst case, when there is vertical motion in the picture at a rate of an odd multiple of one scanning line per second, an entire field's worth of resolution is lost.

To better comprehend this, consider as an extreme example a very thin horizontal line moving vertically through the picture from top to bottom (or from bottom to top) at the frame rate of 1/30 second. This line would move at a rate of two scan lines per frame, so that for the entire time it is in the scanned picture, it would always be located in either an odd-line field or an even-line field, depending on when it entered the scanned picture. The line might, therefore, fall upon each successive scan line in the field that is being scanned when it arrives there. If so, it would appear to flash on and off at the frame rate as it travels through the picture. It is also possible that it could fall upon each successive scan line in the field that is not being scanned when it arrives there. If this happens, it will move completely through the picture without ever being seen at all. The result of this in real television pictures is that the vertical resolution of interlaced pictures varies dynamically between nearly the line count of an entire frame and nearly the line count of a single field, depending on the degree and speed of the vertical motion components in their content.

In 1967, a Bell Laboratories study concluded that the degree of resolution enhancement over the number of lines in a single field that was realized from interlaced scanning depends on picture brightness, but in pictures of normal brightness the enhancement typically amounts not to 100 percent, but to 60 percent. This is an interlace factor (not a Kell factor -- Kell's research was done on progressively scanned images) of 0.6. These findings agree with a previous study that was published in 1958, and with the results of testing done by the Japanese broadcaster NHK in the early 1980s. The NHK study concluded that the picture quality that may be achieved with interlacing is nearly equivalent to that achieved with progressive scanning, using only 60 percent of the scanning lines. We note that tests conducted in the 50s, the 60s, and the 80s all produced virtually the same conclusion. Interlaced scanning was developed as a compromise to fit an adequate number of scanning lines and light flash repetitions into a signal of acceptable analog bandwidth. Different considerations apply in DTV, where the restriction is on digital bit-rate and the quality of digital compression, rather than on analog bandwidth. This has led many to ask if interlaced scanning is still necessary in the world of DTV.