Q: [I believe there is a more accurate explanation for the] 4:1:1 and 4:2:2 color sampling in Bill Davis's article, "DV Demystified," in Videomaker, October 2005.
The first number refers to 4 luminance samples. The second number tells the number of times (out of each 4 luminance samples) that the digital video samples both the red color difference signal and the blue color difference signals in an odd-numbered scan line. The third number tells the number of times (out of each 4 luminance samples) that the digital video samples both color difference signals in an even-numbered scan line. Remember that interlaced video first scans a field that contains lines numbered 1, 3, 5, etc. and then scans a field that contains lines numbered 2, 4, 6, etc. These two fields combine to form one frame of video.
In the particular case of Mini DV's "4:1:1," the two "1" numbers mean that this video samples color one-quarter as often as it samples luminance in the odd scan lines and also in the even scan lines. In the particular case of MPEG-2's "4:2:0," the second number, "2," means that this video samples color half as often as it samples luminance in the odd scan lines, and the third number, "0," means that this video does not sample either color difference signal at all in the even scan lines.
4:1:1 (or DVPRO 50's 4:2:2) is desirable for interlaced video that is transmitted in real time, because it provides color information during both the odd fields and the even fields. However, 4:2:0 is far better for MPEG-2 compression, which operates on de-interlaced video frames. The reason is that, whereas 4:2:2 saves only half the color samples in the horizontal direction, it does no color sub-sampling in the vertical direction. MPEG-2 gets the benefit of a lower data rate, because 4:2:0 saves only half of the color samples in the vertical direction as well as in the horizontal direction.
In the correct definition of 4:1:1 or 4:2:2: sub-sampling, the first number refers to 4 luminance samples. The second number tells the number of times (out of each 4 luminance samples) that the digital video samples both red and blue color difference signals in an odd-numbered scan line. The third number tells the number of times (out of each 4 luminance samples) that the digital video samples both color difference signals in an even-numbered scan line. This definition makes sense of the important "4:2:0" sub-sampling used in MPEG-2, where the "0" means that MPEG-2 performs no sampling of colors in even-numbered scan lines.
A: Thanks for taking the time to write to Videomaker with your concerns regarding the recent article I wrote, titled "DV Demystified."
Your description of the underlying sampling structure is correct on its face. The craft of transmitting fundamentally technical information to a lay audience is often a delicate balance between the technically correct and the readily accessible. In a technical journal or when publishing a "white paper" on a technical subject, such precision is certainly the order of the day.
In a general circulation enthusiast's publication such as Videomaker, adhering to the same level of technical precision can, under certain circumstances, undermine the very mission of the publication -- distributing useful, practical, general information to a lay audience.
The truth of the matter is that the entire field of digital video is rife with technically imprecise shorthand that might help the lay person to grasp the attendant concepts (e.g. "30 frames per second" shorthand when we're actually referring to 29.97 video). I used the shorthand explanations in a context where we were not discussing MPEG-2 nor 4:2:0 sampling. I felt the more concise description would serve my audience well (not to mention help me meet the strict word limit imposed by my editor).
Mathematician Blaise Pascal wrote what is commonly shorthand translated as "I would have written you a shorter letter, if I had more time." Writing when there's no artificial limit on length is relatively easy. It's when you have to be clear, concise (and perhaps even entertaining!) that the real challenge ensues.
"DV Demystified" author and Videomaker Contributing Editor