Storage capacity (for video and film) This is just arithmetic. You can work all these figures out but it’s really useful having some of the key numbers already to hand. Using the ITU-R BT.601 4:2:2 digital coding standard for SD, each picture occupies a large amount of storage space – especially when related to computer storage devices such as DRAM and disks. So much so that the numbers can become confusing unless a few benchmark statistics are remembered. Fortunately the units of mega, giga and tera make it easy to express the vast numbers involved. Storage capacities for SD video can all be worked out directly from the 601 standard. Bearing in mind that sync words and blanking can be re-generated and added at the output, only the active picture area need be stored on disks. In line with the modern trend of many disk drive manufacturers, kilobyte, megabyte and gigabyte are taken here to represent 103, and 109 respectively. Every line of a 625/50 or 525/60 TV picture has 720 luminance (Y) samples and 360 each of two chrominance samples (Cr and Cb), making a total of 1,440 samples per line.

625/50 format There are 576 active lines per picture creating 1440 x 576 = 829,440 pixels per picture. Sampled at 8 bits per pixel (10 bits can also be used) a picture is made up of 6,635,520 bits or 829,440 8-bit bytes – generally written as 830 kB. With 25 pictures a second there are 830 x 25 = 20,750 kbytes or 21 Mbytes per second.

525/60 format There are 480 active lines and so 1,440 x 480 = 691,200 pixels per picture. With each pixel sampled at 8-bit resolution this format creates 5,529,600 bits, or 691.2 kbytes per frame. At 30 frames per second this creates a total of 21,039 kbytes, or 20.7 Mbytes per second. Note that both 625 and 525 line systems require approximately the same amount of storage for a given time – 21 Mbytes for every second. To store one hour takes 76 Gbytes. Looked at another way each gigabyte (GB) of storage will hold 47 seconds of non-compressed video. 10-bit sampling uses 25% more storage. If compression is used, and assuming the sampling structure remains the same, simply divide the numbers by the compression ratio. For example, with 5:1 compression 1 GB will hold 47 x 5 = 235 seconds, and 1 hour takes 76/5 = 18 GB (approx). The storage requirement for VBR compression cannot be precisely calculated but there is usually some target average compression ratio or data rate figure quoted.

Mobile/Wireless/Web All media are limited by the bandwidth available in the transmission/ delivery channel. The most restricted cases are found in wireless and mobile applications where there are a variety of screen sizes, shapes and resolutions ranging from VGA PDAs (480x640) and some 3G phones with up to 320x240, or 176x144 pixels and frame rates down to 15Hz.

HD There are many video formats for HD but the 1080 x 1920 format is popular. Using 4:2:2 sampling, each line has 1920 Y samples and 960 each of Cr and Cb = 3840 samples per line. So each picture has 3840 x 1080 = 4.147 M samples. For 10-bit sampling each picture has the equivalent data of 5.18 M (8-bit) bytes. Assuming 30 pictures (60 fields) per second these produce 155 M bytes/s – 7.4 times that of SD. An hour of storage now needs to accommodate 560 GB.

2K and 4K 2K is a format used in digital film production that uses 4:4:4 10-bit sampling and RGB color space with an image size of 2048 x 1536, and has 24 frames per second. This makes one frame 11.80 MB, and an hour of storage 1.04TB. Note that applied to digital cinema exhibition, the 2K pixel size is 2048 x 1080, and the color space is X´Y´Z´ and uses 12-bit 4:4:4 sampling, as defined by the DCI. The 4K image size is increasingly being used for digital movies. It is a x2 version of 2K, making x4 the pixel area. Here are some popular TV and digital film formats showing the volume of their uncompressed data. Compression of up to 100:1 is applied to MPEG-2 TV transmissions – over 100:1 may be used with more advanced codecs such as MPEG-4 and VC-1. DCI have given a maximum data rate for replay rate for digital cinemas of 250 Mb/s. Here JPEG 2000 compression is used and there is no inter-frame compression; this works out at a compression of about 6.4:1 for 2K and 25.5:1 for 4K. See also: Byte, ITU-R BT.601, ITU-R BT.709, SMPTE 272M

Streaming (video and/or audio) Refers to supplying a constant service, often realtime, of a medium. Although broadcast TV has done this from the beginning and SDI streams data, the term is one more usually associated with delivery by networks, including the Internet. The transmission comprises a stream of data packets which can be viewed/heard as they arrive though are often buffered, stored slightly in advance of viewing, to compensate for any short interruptions of delivery. For the Internet, media is usually highly compressed to offer acceptable results with 28 kb/s for audio and upwards of 64 kb/s for video. There are three predominant video streaming solutions: RealNetworks with RealVideo, RealAudio and RealPlayer, Microsoft Windows Media and Apple Quicktime – each with their particular advantages. As Internet transfers are not deterministic, pictures and sound may not always be constantly delivered. See also: File transfer, Isochronous

Structured Query Language (SQL) A popular language for computer database management. It is very widely used in client/server networks for PCs to access central databases and can be used with a variety of database management packages. It is data-independent and device-independent so users are not concerned with how the data is accessed. As increasing volumes of stored media content are accessible over networks, SQL is able to play a vital role in finding any required items.

Sub-pixel A spatial resolution smaller than that described by a pixel. Although digital images are composed of a matrix of pixels it can be very useful to resolve image detail to smaller than pixel size, i.e. sub-pixel. For example, the data for generating a smooth curve on the screen needs to be created to a finer accuracy than the pixel grid itself – otherwise the curve will look jagged. Again, when tracking an object in a scene, executing a DVE move, or calculating how a macroblock in MPEG-4 AVC coding moves from one picture to another, the size and position of the manipulated picture or element must be calculated, and the picture resolved, to a far finer accuracy than the that of whole pixels – otherwise the move will appear jerky or wrong. See also: Pixel, Tracking

A B C D E F G H I J

K L

M N O P Q R S T U V

89

W X Y Z

GLOSSARY OF TERMS