Video
Non-square Pixels
Pixel shapes
All computer displays and graphic formats use square pixels. What does this mean? Simply that each pixel represents the same distance horizontally as vertically. The number of pixels/inch (pel/in) may be almost any value you like, but if an image is, say, 100 pel/in, then it is 100 pel/in in both directions.
This is a fairly obvious way to do things. If you were to have one value of pel/in horizontally and another one vertically it's clear that image manipulation could get quite tricky. Yet this is exactly the situation with several video formats, especially the two that are becoming the most popular - DV and MPEG. In these formats, the pixels are not square but rectangular, meaning the horizontal and vertical pel/in values are different.
Why?
Why should anyone want to use rectangular pixels? The root of the issue is the fact that there are two systems for TV around the world, although everyone would really prefer there to be one.
- The European-based systems run at 25 frames/second, 576 video lines/frame
- The American-based systems run at 30 frames/second, 480 video lines/frame
The terms NTSC, PAL, & SECAM actually refer to the colour encoding technique, not the frame rate, although in practice all 30 fps systems use NTSC and 25 fps system use PAL or SECAM
The reason for the different frame rates is historic, based on the frequency of the local mains supply (50 or 60 Hz). Before the advent of stable solid-state electronics, broadcast TV had to use a field rate (2 fields/frame) that was the same as the mains supply frequency to avoid visible interference. The technical problem disappeared more than 25 years ago, but backward compatibility of consumer equipment has meant the continuing distinction.
A quick bit of arithmetic on the two systems reveals an interesting connection: they both add up to 14,400 video lines/second. That was by design, not coincidence.
Data rate
An anlogue video picture is actually only analogue horizontally. Vertically it is already divided into a fixed number of lines. The resolution that a TV signal is capable of depends on how rapidly the signal can change as the line scans across the screen. To digitise a TV picture the signal has to be sampled as it scans the screen. The ideal sampling rate should reflect the inherent resolution of the analogue signal.
In both TV systems the number of lines/second is the same, and each line carries the same amount of information. It is therefore logical to sample TV lines at the same rate for both systems. It also means that the sampling process yields the same number of bits/second in the resulting data. For digital signal processing purposes this minimises the difference between the systems.
This is exactly how the international standards for digital video and TV were drawn up several years ago. It was decided that a digital TV picture line would consist of 720 pixels. This is fixed for both 25 & 30 fps systems. (The technically minded might like to know that both systems sample at exactly 13.5 million pixel per second.)
However, the number of vertical pixels that make up one frame has to be equal to the number of TV lines, and is different between the systems, being either 576 or 480.
This means that all video frame formats derived from these standards are 720 pixels wide: 720 x 576 for 25 fps and 720 x 480 for 30 fps. This includes digital broadcast and DVD, as well as professional and consumer DV camcorders formats.
Site sections:
