Making sense of 4:4:4, 4:2:2 and 4:2:0

I was looking for more information about 4:2:0 and 4:2:2 as referenced in the July article by William Zou on light compression. I have been unable to
Publish date:
Social count:

I was looking for more information about 4:2:0 and 4:2:2 as referenced in the July article by William Zou on light compression. I have been unable to find a definitive source and believe that an exact definition would help with some of my confusion in understanding this article. Any help would be appreciated. Thanks,

Kevin A. McGrail You are not the only one confused by that nomenclature. As a matter of fact, the signals represented are not entirely consistent across several very similar nomenclatures. To understand these, an understanding of signal sampling is needed. Based on Nyquist theory, a signal must be sampled at a rate at least double the highest frequency found within the signal to avoid aliasing. In other words, if you want to sample a 20kHz signal, the sampling rate must be at least 40kHz. Higher sampling rates are generally preferred, but like anything else, higher rates come at a price. In this case, the price is an increase in the amount of data generated.

In the early days of digital video, a sampling rate of four times the color subcarrier (3.58MHz X 4 = 14.3MHz) was used to sample composite NTSC video. A different sampling rate was used for PAL. Later, it was agreed that video would be sampled using 13.5MHz. Even though that value is not four times any specific frequency used in NTSC or PAL, it is common to both standards and works well as a standard sampling frequency.

With that said, a 4:4:4 signal could be digital representation of a component analog RGB signal that was sampled at 13.5MHz. A 4:4:4 signal could also be a Y, R-Y, B-Y signal that was sampled at 13.5MHz. The "4" represents a signal sampled at 13.5MHz. Likewise, a "2" is used to represent a signal sampled using a sample rate of 6.75MHz (one-half of 13.5Hz). A "1" is used for signals sampled at a rate of 3.375MHz (one-fourth of 13.5MHz). Figure 1 shows graphically where the samples are taken based on the nomenclature used.

As stated, various nomenclatures are used. Generally, three numbers separated by colons depict the sampling rate. If a fourth number is used it normally refers to an alpha or key channel. The first number normally refers to the luminance channel, while the next two numbers are for the color difference channels. However, variations exist, such as using 4:4:4 to represent RGB or using 4:2:0 to represent the fact that two samples are taken on every other line. HD signals have thrown another wrench in the works, as manufacturers are using 4:2:2 to represent a "normal" HD signal. These numbers are based on the "4" representing 74.25MHz. Relative to the original system ("1" = 3.375MHz), these HD numbers are actually the equivalent of 22:11:11.

Readers: We regularly receive requests for information on maintaining obsolete equipment, especially transmitters. The issue of parts, documentation and support are foremost in most of these readers' minds. So the call is going out for help.

Do you have resources on parts, manuals and support for NEC and Townsend transmitters? If you have contacts that may be helpful to other engineers who have these units, parts or manuals for sale, forward the information to us and we'll make it available to others. Send the information to: