DIGITAL

The future of television is digital.... not really because it's "better", but because it just is. There are many new concepts and gizmos, and no shortage of new TLA's (three-letter-acronyms). Digital signal handling has many benefits though, not the least of which is that it's usually cheaper, even if it doesn't seem that way at first glance. Here are some features in no particular order:

• There is no picture degradation except when converting from one format to another.
• Computer stuff is "cheap like borscht". Many new features are easily implemented like multiple aspect ratios, frame rates and resolutions.
• The audio is better.
• DVDs with their interactive menus allow us to present information in a non-linear fashion. DVDs are a much better home playback system than affordable analog tape can ever be.
• Digital video (stockshot libraries, news stories, etc.) can be distributed to every desktop in a workplace if a fast network is in place.

Analog electronics get very expensive when dealing with high-frequency, low-noise, low-distortion signals. Digital electronics works with signals that are either "on" or "off", making the electronic components very "cheap and dirty". As long as the base two digits can still be discerned, it doesn't matter how much noise and interference is mixed into them.

Basic "Digit"izing

To digitize an analog "squiggly voltage" it is sampled at known intervals, and it's amplitude at that point is assigned a number. If enough samples are taken, and if the number of values in the scale are sufficient, the analog signal can be re-created faithfully.

• Video is usually sampled at about 14 million samples per second, with each sample having the range of an eight-bit byte (256 levels). 
• Audio for production is sampled at  48,000 samples per second, with sixteen bits holding the value (two bytes or 65,536 levels).
• CD audio is sampled at  41,100 samples per second, with sixteen bits holding the value (two bytes or 65,536 levels).

 4:2:2      4:1:1     4:1:0... Hike!

Digital television still uses the the colour matrix idea (Y, R-Y. B-Y) and digitizes the three component signals as described above, but, to "save bits", the sample rates often aren't the same for the three components. Recall that the Y channel holds all of the picture details. Different sampling schemes are described by their sample rate ratios, 4:2:2, etc. Here is the story (don't shoot the messenger!)

Digitizing video began with the first digital timebase correctors (tbc's). At first, the entire composite signal, subcarrier and all was sampled. The sampling rate needed to be a multiple of the subcarrier, and "four times subcarrier" became standardized. That's 14.318180 megahertz, in case you forgot... 14 million-and-change samples-per-second.

When Betacam and MII introduced component video recording (Y/ R-Y/ B-Y) there was no subcarrier to deal with, and the sampling frequency was changed to 13.5 megahertz, which works out to 720 samples per active line in both NTSC and PAL. Handy. Don't try to do the math... the active part of the line is the visible part without taking blanking or sync into the equation. So. There's the 4 in 4:2:2... it's four times a number that we don't use any more! 

Let's go after the 2's now. Analog Betacam (and MII) used pairs of rotating heads that drew two tracks at once. One track contained the Y signal, and the other contained all of the colour in an interesting way. The two colour components were squashed to half width, and recorded one after the other. This resulted in the two colour components getting only half the number of samples that the Y channel gets... two for four, or a Y: R-Y : B-Y ratio of 4:2:2. Having half of the samples doesn't leave gaps in the picture, rather, the pixels are twice the width (half the detail).

 But Wait... there's More!

4:1:1 and the Rest of 'em

A digital video recorder doesn't record video, audio, or signals of any kind... it records BITS. Less BITS means more time on the tape, or a smaller tape. A digital format that records 4:1:1 only records one colour component sample for every four in the luminance. This is not as grotty as it sounds because it is colour detail that is lost. The fine picture detail in the Y channel is still sampled at a high rate. There are many ways of coding and recording video and audio.... one high-quality system actually digitizes the composite signal, subcarrier and all! Most systems use the Y channel / colour component idea in one form or another. 

If a signal contains a full-quality "alpha" channel (used to control the blending with other images) it might be quoted as 4:2:2:4, some high-end equipment uses 4:4:4:4, and a very high-falutin' production house can even choose 8:8:8:8!

More often, we run across lower-quality systems, like 4:1:1 (DVCam, for instance). There is even a 4:1:0... this doesn't mean that there is no blue in the picture, rather, each of the colour components is sampled one-quarter as often... that's the 4:1 part of the ratio. The zero at the end indicates that something isn't sampled, and that is true. On the first line, only ONE of the colour components is sampled, and on the next, the OTHER component is sampled. When the picture is re-constructed, the missing samples are filled in by the samples from the line above. This lowers the vertical resolution of detail in the colour channels by half.