Tag : Digital Basics

1.125 MHz

This is a common frequency that can be derived from 625/50 PAL (and 525/60 NTSC) that runs through SD, HD and UHD digital television standards using 4:4:4 or 4:2:2 sampling, including 1080-line HD and UHD at 25, 30, 60 Hz frame rates. It can be created from the old analog (PAL or NTSC) black-and-burst signal. Because 1.125 MHz is a common frequency, black and burst was widely used as a timing reference signal in the early days of HD.

See also: 13.5, Tri-level sync

10-bit lin

A type of digital sampling of analog images that creates 10-bit (210, 1024 possible levels) numbers to describe the analog brightness levels of an image. Lin, short for ‘linear’ means the levels are assigned evenly to the levels of the analog signal they describe. So an LSB change describes the same change in level if it is in a bright area or a dark area of the picture. Most professional HD and some SD television is sampled this way according to ITU-R BT.601 and 709. 10-bit lin sampling allows good quality to be maintained through TV production and post production where the processes can make particular demands outside the range of normal viewing, and so produces good results for viewers. However if color grading is required then the wide dynamic range that can be described by 10-bit log would be more useful, or indeed one of the newer high dynamic range formats.

See also: 10-bit log, Gamma, Color Space

10-bit log

This usually refers to a 10-bit sampling system that maps analog values logarithmically rather than linearly. It is widely used when scanning film images which are themselves a logarithmic representation of the film’s exposure. This form of sampling is available directly from some digital cinematography cameras.

The 10-bit data can describe 210 or 1024 discrete numbers, or levels: 0 – 1023 for each of the red, blue and green (RGB) planes of an image. However, as all electronic light sensors have a linear response and so produce an output directly proportional to the light they see, when scanning film they represent the transmittance of the film. Usually it is negative film that is scanned and this means a large portion of the numbers generated describe the scene’s black and dark areas (representing bright areas of the original scene), and too few are left for the light areas (dark scene) where ‘banding’ could be a problem – especially after digital processing such as grading and color correction. Transforming the numbers into log (by use of a LUT) gives a better distribution of the digital detail between dark and light areas and so offers good rendition over the whole brightness range without having to use more bits. A minimum of 13-bit linear sampling converted to 10-bit log sampling means sufficient detail in the pictures is stored to allow headroom for downstream grading that is common in film production.

10-bit log is the basis for sampling in the Cineon and SMPTE DPX formats that are still widely used in the post production and DI industries.

See also: 10-bit lin, LUT


Historically the nominal 30 frames/60 fields per second of NTSC color television is usually multiplied by 1000/1001 (= 0.999) to produce slightly reduced rates of 29.97 and 59.94 Hz. This offset gives rise to niceties such as drop-frame timecode (dropping one frame per thousand: 33.3 seconds) and audio that also has to run in step with the video. Although having strictly analog origins, dating from the very beginning of NTSC color transmissions in 1953 as a fix-it to avoid a clash of frequencies, the 1000/1001 offset has been extended into the digital, HD and UHD world where 24 Hz becomes 23.976 and 30 frames/60 fields per second are again changed to 29.97 and 59.94 Hz. Of course, as the frame/field frequency changes, so do the line and color subcarrier frequency as they all have to be locked together. Note that this does not apply to PAL color systems as these always use the nominal values (25 Hz frame rate).

The reason for the 1000/1001 offset is based in monochrome legacy. Back in 1953, the NTSC color subcarrier was specified to be half an odd multiple (455) of line frequency to minimize the visibility of the subcarrier on the picture. Then, to minimize the beats between this and the sound carrier, the latter was to be half an even multiple of line frequency, and to ensure compatibility with the millions of existing monochrome TV sets, the sound carrier was kept unchanged – at 4.5 MHz – close to 286 times the line frequency (Fl). Then, in a real tail-wags-dog episode, it was decided to make this exactly 286 times… by slightly altering the line frequency of the color system (and hence that of the color subcarrier and frame rate). Interestingly it is said that the problem could soon have been solved with a little improved engineering, so avoiding the need for this awkward frequency offset and all the many thousands of hours of additional engineering and operational effort this has caused down the years.

Here’s the math.

Fl = frames per second x number of lines per frame
Nominally this is: 30 x 525   = 15,750 kHz

But it was decided that: 286 x Fl     = 4.5 MHz
So:                                     Fl = 4,500,000/286          = 15,734.265 kHz
This reduced Fl by:         15734.265/15750           = 1000/1001 or 0.999

As all frequencies in the color system have to be in proportion to each other, this has made:

NTSC subcarrier (Fl x 455/2) = 3.579 MHz
30 Hz frame rate (Fl/number of lines per frame) = 29.97 Hz

Following on, all digital sampling locked to video is affected so, for example, nominal 48 and 44.1 kHz embedded audio sampling becomes 47.952 and 44.056 kHz respectively.

As the reasons for ‘drop-frame’ were analog, it is not a requirement of digital television but still the frequencies appear in digital TV standards, and they are widely used, even though analog TV transmissions are now switched off in most countries.

See also: Drop-frame timecode, Embedded audio

13.5 MHz

This is the sampling frequency of luminance in SD digital television systems as defined by the ITU. It is represented by the 4 in 4:2:2. The use of the number 4 is pure nostalgia as 13.5 MHz is in the region of 14.3 MHz, the sampling rate of 4 x NTSC color subcarrier (3.58 MHz), used at the very genesis of digital television equipment.

Reasons for the choice of 13.5 MHz belong to politics, physics and legacy. Politically it had to be global and to work for both 525/60 (NTSC) and 625/50 (PAL) systems. The physics is the easy part; it had to be significantly above the Nyquist frequency so that the highest luminance frequency, 5.5 MHz for 625-line PAL systems, could be faithfully reproduced from the sampled digits i.e. sampling in excess of 11 MHz but not so high as to produce unnecessary, wasteful amounts of data. Some math is required to understand the legacy.

The sampling frequency had to produce a static pattern on both 525 and 625-line standards, otherwise it would be very complicated to handle and, possibly, restrictive in use. In other words, the frequency must be a whole multiple of the line frequencies of both standards.

The line frequency of the 625/50 system is simply: 625 x 25 = 15,625 Hz
(NB 50 fields/s makes 25 frames/s)
So line length is 1/15,625 = 0.000064 or 64µs

The line frequency of the 525/60 NTSC system is complicated by its offset factor of 1000/1001 to avoid interference when transmitted. The line frequency is 525 x 30 x 1000/1001 = 15,734.265 Hz. This makes line length 1/15,734.265 = 63.5555µs

The difference between the two line lengths is 64 – 63.55555 = 0.4444µs

This time divides into 64µs exactly 144 times, and into 63.5555µs exactly 143 times. This means the lowest common frequency that would create a static pattern on both standards is 1/0.4444 MHz, or 2.25 MHz.

Now, back to the physics. The sampling frequency has to be well above 11 MHz, so 11.25 MHz (5 x 2.25) is not enough. 6 x 2.25 gives the adopting sampling frequency for luminance of 13.5 MHz.

Similar arguments have been applied to the derivation of sampling for HD. Here 74.25 MHz (33 x 2.25) is used for luminance sampling.

See also: 1.125 MHz4:1:1, 4:2:0, 4:2:2, 4:4:4, 4fsc, Nyquist (frequency)

2.25 MHz

This is a common frequency that can be derived from 625/50 PAL (and 525/60 NTSC) that runs through SD, HD and UHD digital television standards. It can be created from the old analog black and burst signal. 2.25 MHz, or multiples thereof, runs through all the major digital television standards. These can be locked to black and burst. They include 1080-line HD at 25, 30 fps.

See also: 13.5, Tri-level sync

2:2 F1/F2

A film frame being transported as 2:2 (pSF) is placed into two consecutive video fields. F1/F2 denotes that the film frame is carried in field one and the following field two. This is commonly referred to “normal dominance” or “perfect cadence”.

See also pSF

2:2 F2/F1

A film frame being transported as 2:2 (psf) is placed into two consecutive video fields. F2/F1 denotes that the film frame is carried in a field two and the following field one. This is commonly referred to “reverse dominance” or “reverse cadence”.

See also pSF

ADC or A/D

Analog to Digital Conversion (or Converter). Also referred to as digitization or quantization. The conversion of analog signals into digital data, normally for subsequent use in digital equipment. For TV, samples of audio and video are taken, the accuracy of the process depending on both the sampling frequency and the resolution of the analog amplitude information: how many bits are used to describe the analog levels. For TV pictures with modern cameras 14 bits or higher is normally used behind the image sensor; for sound, 16, 20 or 24 bits are common. The ITU-R BT.601 standard defines the sampling of SD video components based on 13.5 MHz, and AES/EBU defines sampling of 44.1 (used for CDs) and 48 kHz for audio. For pictures the samples are called pixels, which contain data for brightness and color. HD and UHD video formats use higher sampling rates and generally more bit depth.

See also: AES/EBU, Binary, Bit, Into digits (Tutorial 1), Pixel


Undesirable ‘beating’ effects caused by the presence of frequencies that are too high in an analog input signal that is converted into digits. Passing the input through a suitable low-pass filter, removing all the frequencies above half that of the analog-to-digital converter’s (ADC) clock rate (the Nyquist Frequency), solves the problem. Examples of aliasing include:

1) Temporal aliasing – e.g. wagon wheel spokes apparently reversing, also movement judder seen in the output of standards converters with insufficient temporal filtering.

2) Raster scan aliasing – twinkling effects on sharp boundaries such as horizontal lines. Due to insufficient filtering this vertical aliasing, and its horizontal equivalent, are often seen on the output of lower quality video processing equipment, such as poor DVEs, as detailed images are re-sized.

The appearance of ‘steppiness’ or ‘jaggies’ of poorly filtered images with near horizontal lines in a TV image is also referred to as aliasing.

See also: Anti-aliasing, Interpolation (temporal) Interpolation (spatial), Into digits (Tutorial 1), Nyquist (frequency)


Techniques to smooth aliasing effects created by poor filtering and other techniques. Perhaps some videos or crude character generation may produce poor, aliased, pictures. Anti-aliasing may then be applied to reduce the effect and improve the look of the image. A better approach is to avoid aliasing in the first place and with good modern technology easily available, serious aliasing should be avoided.

See also: Aliasing, Interpolation (spatial), Interpolation (temporal)

Areal density

The density of data held on an area of the a recording medium’s surface. This is one of the parameters that manufacturers of disk drives and tape recorders constantly strive to increase. For example one currently-available high-capacity drive from Seagate achieves around 1Tb/square inch. When compared to the 130 Gb figure reported here in the 2008 edition of this book, this shows not only the continuing high rate of development of disk-drive technology, but also that seven years is too long between editions of the Digital Fact Book! With development continuing apace, yet greater capacities can be expected in future editions.

See also: Hard disk drives

Website: www.seagate.com


Average Revenue Per Unit, usually used by telecoms companies, to describe the money made from each ‘unit’ or ‘customer’!


Particular visible effects on images which are a direct result of some technical limitation. Artifacts are generally not described by traditional methods of signal evaluation. For instance, the visual perception of contouring in a picture cannot be described by a signal-to-noise ratio or linearity measurement.


Application Specific Integrated Circuit. Custom-designed integrated circuit with functions specifically tailored to an application. These replace the many discrete devices that could otherwise do the job but work up to ten times faster with reduced power consumption and increased reliability. ASICs are now only viable for very large-scale high volume products due to high start-up costs and their inflexibility as other programmable devices, such as FPGAs (field programmable gate arrays), offer more flexible and cheaper opportunities for small-to-medium sized production levels.

See also: PLD

Aspect Ratio

1. of pictures. The ratio of length to height of pictures. All TV screens used to be 4:3, i.e. four units across to three units in height, but now all new models are widescreen, 16:9. Pictures presented this way are believed to absorb more of our attention and have obvious advantages in certain productions, such as sport. In the change towards 16:9 some in-between ratios were used for transmission, such as 14:9.

2. of pixels. The aspect ratio of the area of a picture described by one pixel. The ITU-R BT.601 digital coding standard for SD defines luminance pixels which are not square. In the 525/60 format there are 486 active lines each with 720 samples of which only 711 may be viewable due to blanking. Therefore the pixel aspect ratios on 4:3 and 16:9 screens are:

486/711 x 4/3 = 0.911 (tall)
487/711 x 16/9 = 1.218 (wide)

For the 625/50 formats there are 576 active lines each with 720 samples of which 702 are viewable so the pixel aspect ratios are:

576/702 x 4/3 = 1.094 (wide)
576/702 x 16/9 = 1.458 (wider)

All HD digital image standards define square pixels.

Account must be taken of pixel aspect ratios when, for example, executing DVE moves such as rotating a circle. The circle must always remain circular and not become elliptical. Another area where pixel aspect ratio is important is in the movement of images between platforms, such as computers and television systems. Computers generally use square pixels so their aspect ratio should be adjusted for SD television-based applications.

See also: ARC, Pixel


The amount of information (data) that can be passed in a given time. In television a large bandwidth is needed to show sharp picture detail in realtime, and so is a factor in the quality of recorded and transmitted images. For example, ITU-R BT.601 and SMPTE RP 125 allow analog luminance bandwidth of 5.5 MHz and chrominance bandwidth of 2.75 MHz for standard definition video. 1080-line HD has a luminance bandwidth of 30 MHz (ITU-R BT.709).

Digital image systems generally require large bandwidths hence the reason why many storage and transmission systems resort to compression techniques to accommodate the signal.

Bayer filter/mask array


A Bayer filter is a matrix of red, green and blue non co-sited filters placed onto an imaging chip (CCD, CMOS) so that it can capture the separate red, blue and green primary colors of the image to record a color digital image. This greatly simplified the construction of color cameras, and somewhat mimics how our (single-retina) eyes see color. As our eyes have more resolution for green light than red or blue, the Bayer filter on the imaging chip has twice as many green cells as red and blue. Some redundancy of the green pixels produces an image which is less noisy and has finer detail than would be achieved if there were an equal number of red, green and blue cells.

For further use, the R, G and B pixels generated by the Bayer-filter-and-imaging-chip combination need to be ‘unmasked’ using a complex algorithm. This process, sometimes called Debayering, produces the separate red, green and blue images that together make up the color image.

Traditionally professional TV cameras have used three image sensors, one to pick up each primary color. This arrangement requires that behind the lens there is a three-way light-splitting glass block delivering the separate R, G and B images to three light sensors that must be accurately registered together. This has involved a considerably more bulky construction and greater cost than is needed by those based on a single image-sensor chip complete with a Bayer filter.

The Bayer filter was patented in 1976 and early use was with consumer stills and video cameras. A number of other variants of RGB filter array are now in use. It was 30 years later that single-chip cameras started to be accepted for the professional video and movie markets, some years after they had been widely used in high-end stills cameras, for example. ARRI Alexa, Canon, Sony F65/55 and RED.



Mathematical representation of numbers to base 2, i.e. with only two states, 1 and 0; on and off; or high and low. This is the basis of the mathematics used in digital systems and computing. Binary representation requires a greater number of digits than the base 10, or decimal, system most of us commonly use everyday. For example, the base 10 number 254 is 11111110 in binary.

There are important characteristics which determine good digital video equipment design. For example, the result of a binary multiplication contains the sum of the number of digits of the original numbers.

10101111 x 11010100 = 1001000011101100
(in decimal 175 x 212 = 37,100)

Each digit is known as a bit. This example multiplies two 8-bit numbers and so the result is a 16-bit number. So, for full accuracy, all the resulting bits should be taken into account. Multiplication is a very common process in digital television equipment (e.g. keying, mixes and dissolves) so exactly how such a result is treated to connect with downstream digital products with 8, 10 or 12-bit digital inputs, raises some interesting questions!

See also: Bit, Byte, Digital mixing, Dynamic Rounding

Bit (b)

Binary digIT = bit

One mathematical bit can define two levels or states, on or off, black or white, 0 or 1 etc.; two bits can define four levels, three bits eight, and so on: generally 2n, where n = the number of bits. In image terms 10 bits can be used to define 1024 levels of brightness from black to white (with ITU-R BT.601 and 709, 64 = black and 940 = white).
Note that in both decimal and binary numbers the first digit describes the largest part of the number’s value. For example, the base-10 number 254 is 11111110 in binary. In binary the first digit of the number is called the most significant bit (MSB). Likewise the last digit is the least significant bit (LSB).

See also: Byte


Burnt-in Timecode. Timecode that is displayed on the video to which it refers. This is often recorded to provide precise frame references for those viewing on platforms not supplied with timecode readers.


Informal word used to describe when streaming media suddenly ‘hits the buffers’ – stops. This is usually due to a lack of bandwidth when viewing video over the internet. In recent years the implementation of adaptive streaming schemes, as well as faster internet, have together greatly reduced the occurrence of buffering, making the viewing of video delivered via the internet a non-buffered experience.


An error in a computer program that causes the system to behave erratically, incorrectly or to stop altogether. Term dates from the original computers with tubes and relays, where real live bugs were attracted by the heat and light and used to get between the relay contacts.

Byte (B), kilobyte (kB), megabyte (MB), gigabyte (GB), terabyte (TB) and petabyte (PB)

1 Byte (B) = 8 bits (b) which can describe 256 discrete values (brightness, color, etc.).

Traditionally, just as computer-folk like to start counting from zero, they also ascribe 2 raised to the power 10, 20, 30, etc. (210, 220, 230, etc.) to the values kilo, mega, giga, etc. which become, 1,024, 1,048,576, 1,073,741,824, etc. To be factually correct, these powers of 2 should actually be called kibibyte (KiB), mebibyte (MiB), gibibyte (Gib), tebibyte (TiB) etc., though this is not very commonly seen in practice.

This can be difficult to handle for those drilled only in base-10 mathematics. Fortunately, disk drive manufacturers, who have to deal in increasingly vast numbers, describe their storage capacity in powers of 10, so a 100 GB drive has 100,000,000,000 bytes capacity. Observation suggests both systems are continuing in use…which could lead to some confusion.

TraditionalNewApprox Duration 
1 kB = 210 bytes = 1,024 B103 B2/3 line1/5 line1/8 line
1 MB = 220 bytes = 1,048,576 B 106 B1 frame1/5 frame130 lines
1 GB = 230 bytes = 1.074 x 109 B109 B47 sec6.4 sec3.5 sec
1 TB = 240 bytes = 1.099 x 1012 B1012 B13¼ hrs1¾ hrs58 mins
1 PB = 250 bytes = 1.126 x 1015 B1015 B550 days74 days40 days

Currently 3.5-inch hard disk drives store from about 100 GB to 6 TB. Solid-state store chips, RAMs, increment fourfold in capacity every generation now offering up to 8Gb chips (i.e. 8 x 230). Flash memory is now widely used in many professional and consumer video cameras.

A full frame of standard definition digital television, sampled at 10 bits according to ITU-R BT.601, requires around 1 MB of storage (1.037 MB for 576-line, 876 kB for 480-line systems). HDTV frames comprise up to 5 or 6 times more data, and 2K digital film frames sampled in RGB or X´Y´Z´ (DCI colorspace) are about 12 MB. 4K quadruples that to 48 MB. Storing the larger formats as uncompressed or applying ‘lossless’ compression makes huge demands on resources.

See also: DCI, DRAM, Disk drives, Storage



Charge Coupled Device (CCD), either assembled as a linear or two-dimensional array of light sensitive elements. Light is converted to an electrical charge in a linear fashion – proportional to the brightness impinging on each cell. The cells are coupled to a scanning system which, after analog to digital conversion, presents the image as a series of binary digits.

Typically the charge created on each cell is shuffled along to the end of its line where it is measured by an ADC that outputs a digital number corresponding to the charge, and hence, the brightness. This handling of the charge exposes it to noise and early CCD arrays were unable to work over a wide range of brightness. Now they offer low noise, high resolution imaging for television and digital cinematography.

See also: CMOS


A simple check value of a block of data intended to recognize when data bits are wrongly presented. It is calculated by adding all the bytes in a block. It is fairly easily fooled by typical errors in data transmission systems so that, for most applications, a more sophisticated system such as CRC is preferred.

See also: CRC

Chroma keying

The process of overlaying one video picture, or part of, over another. The areas of overlay are defined by a specific range of color, or chrominance, on the background video signal that is used to create a key signal for a chroma keyer. For this to work reliably, the chrominance must have sufficient resolution, or bandwidth. PAL or NTSC analog coding systems significantly restrict chroma bandwidth and so are of very limited use for making a chroma key which, for many years, was restricted to using live, RGB camera feeds.

An objective of the ITU-R BT.601 and 709 digital sampling standards was to allow high quality chroma keying in post production. The 4:2:2 sampling system allows far greater bandwidth for chroma than PAL or NTSC and helped chroma keying, and the whole business of layering, to thrive in post production. High signal quality is still important to derive good keys so some high-end operations favor using RGB (4:4:4) for keying – despite the additional storage requirements. Certainly anything but very mild compression tends to result in keying errors appearing – especially at DCT block boundaries.

Chroma keying techniques have continued to advance and use many refinements, to the point where totally convincing composite images can be easily created. You can no longer see the join and it may no longer be possible to distinguish between what is real and what is keyed.

See also: Color space, Digital keying, Photo-real


The color part of a television signal, relating to the hue and saturation but not to the brightness (luminance) of the signal. Thus pure black, gray and white have no chrominance, but any colored signal has both chrominance and luminance. Although imaging equipment registers red, blue and green, television pictures are handled and transmitted as U and V, Cr and Cb, or (R-Y) and (B-Y), which all represent the chrominance information of a signal, and the pure luminance (Y).

See also: YUV, Y,Cr,Cb, Composite


An exact copy, indistinguishable from the original. As in copying recorded material, eg copy of a non-compressed recording to another non-compressed recording. If attempting to clone compressed material care must be taken not to decompress it as part of the process or the result will not be a clone. Today, cloning is best done by simply copying a file.


Complementary Metal-Oxide Semiconductor technology is very widely used to manufacture a wide range of electronic integrated circuits (chips). CMOS chip digital applications include microprocessors, RAM and dynamic and static memory. They also have a variety of analog applications.

CMOS devices are favored for their immunity to high levels of noise, low static power drain, with significant power only drawn while the transistors switch, and high density packing of logic functions. Being so widely used, the technology is relatively cheap to manufacture.

CMOS imaging sensors are potentially cheaper to make than the alternative CCDs. They also consume less power, can be more light sensitive (and so faster, less noisy and better in low lights), have less image lag and can include image-processing functions on each photosite (cell) on the chip. Each photocell can have its own ADC, so the charge created by the light falling on it is converted into digits on site, then the data is passed to the highways. This way, CMOS imaging sensors with their on-site digitizing, have a much lower noise level and so can work in lower light conditions than CCDs.

CMOS technology also plays a vital role in digital projection where DMD (Digital Micromirror Device) chips make images from data.

See also: DMD

Co-sited sampling

This is a sampling technique applied to color difference component video signals (Y, Cr, Cb) where the color difference signals, Cr and Cb, are sampled at a factor of the luminance, Y, frequency – for example as in 4:2:2. If co-sited sampling is applied, the two color difference signals are sampled at the same instant, and simultaneously with a luminance sample. Co-sited sampling is the ‘norm’ for component video as it ensures the luminance and the chrominance digital information is coincident, minimizing chroma/luma delay.

4:2:2 Co-sited sampling

Co-sited sampling



Originally short for a combination of a coder and decoder but now often used to describe just one or the other. Mostly codec refers to a compression coder or decoder such as JPEG, MPEG or JPEG 2000.

Color cube

A representation of color space by a three-dimensional diagram. For example, all definable colors of an RGB color space can be contained in an RGB color cube where R, G and B are axes at right angles to each other (like x, y and z at the corner of a cube). Different color spaces and interpretations of color are defined by different color cubes.

If the exact spectral values of R, G and B are defined, that cube defines an absolute color space. Such cubes are available from a number of vendors.

Color science

Color Science has been an area of scientific research for over 100 years. Brought together in 1931 by the CIE (Commission Internationale de l’Eclairage) this sphere of science studies all aspects of human perception of color and brightness. Early use was to study how dyes could be mixed. The issues of color printing and the effects of different viewing conditions on perception. There are large amounts of research and books published relating to this subject.

The definitions for all our television and cinema viewing standards are rooted in color science. The numeric definition of r,g,b and the conversions to YCrCb are examples of the practical use of color science. In this multimedia world media transportability and maintenance of creative intent would not be possible without color science defining the solid core of math to support the industry.

See also: Color Space

Color space

The color range between specified references. Typically three reference points are quoted as in television is could be RGB, Y R-Y B-Y, or Hue, Saturation and Luminance (HSL). These are all color spaces. In print, typically Cyan, Magenta, Yellow and Black (CMYK) are used. Film is RGB while digital cinema uses X´Y´Z´. Pictures can be moved between these color spaces but it requires careful attention to the accuracy of the involved processing. Operating across the media – in print, film and TV, as well as between computers and television – requires color-space conversions to display the correct colors everywhere.

Electronic light sensors detect red, blue and green (RGB) light but TV signals are usually changed into Y, R-Y and B-Y components as, or very soon after they enter the electronic realm via camera, scanner or telecine. There is some discussion about which color space is best for post production – the most critical operation being keying. However, with most video storage and infrastructure being component-based, the full RGB signal is usually not available, so any of its advantages can be hard to realize for television-based productions. However, in the Digital Intermediate (DI) process, where movie footage undergoes ‘post production’, RGB color space predominates.

With the increasing use of disk storage, networking able to carry RGB and digital cameras with RGB outputs, RGB infrastructure and operations are more widely used. Even so, RGB takes up 50 percent more storage and, for most productions, its benefits over component working are rarely noticed. One area that is fixed on RGB use is in 2K and 4K digital film (digital intermediates). Modern digital techniques allow the use of both RGB and Y R-Y B-Y to best suit the requirements of production as well as those of distribution and delivery.

More recently the world has become more complex for color, with a wide diversity of camera, display systems and methodologies for moving from one to the other. The CRT display set the colorimetric standard that still holds today even though CRTs are no longer used and other display technologies, LCD, plasma, micro-mirror and OLED, for example, each have their own characteristics for color and light output transforms.

In the television broadest description Color Space defines how all parts of the imaging chain respond to light or electrical drive. Cameras separate out the red, green and blue light using filters. The spectral response, in particular the passable bandwidth, of these filters control the margins of the range of colors that can be represented. These are usually represented in a diagram of the type shown below.

Color Space

The axes of this diagram are the xy of the color notation xyY where the Y component would represent the intensity. The human visible area is infilled with color and, on more detailed depictions of this diagram, show the wavelengths of light representing each color. Saturation increases towards the edge of the locus.

The corners of the triangles forming the ACES, DCI P3 and Rec.709 are at the coordinates of the primary colors used in these displays. Only colors inside the respective triangle can be reproduced by that display. You can see that no display technology covers the full human range for color and that, for example, full green on a Rec.709 monitor is not the same color as full green on the P3 display. Some colors in nature cannot be represented at their proper saturation using today’s imaging technology.

See also: 2K, Keying

Color Transformation Language (CTL)

Color Transformation Language is a small programming language designed to serve as a building block for digital color management systems. It allows users to describe color transforms in a concise and unambiguous way by expressing them as programs that describe the transform that can be applied to pixel data. It is designed to run fast, operating on many pixels at one time.

See also: OpenEXR

Website: http://ampasctl.sourceforge.net


The linking together of systems in a linear manner. In digital television this often refers to the concatenation of compression systems which is a subject of concern because any compression beyond about 2:1 generally results in the removal of information that cannot be recovered. As the use of compression increases, so too does the likelihood that material will undergo a number of compressions between acquisition and transmission. Although the effects of one compression cycle might not be very noticeable, the impact of multiple decompressions and recompressions (A.K.A. decoding and recoding), with the material returned to baseband in between, can cause considerable damage. The damage is likely to be greatest where different compression schemes are concatenated in a particular signal path.



An unwanted artifact, similar to posterization. Digital video systems exhibit contouring when an insufficient number of quantizing levels are used, or inappropriate processes are used such as poor truncation – rounding of LSB errors. The result is that the picture’s brightness changes in too large steps and become visible over relatively even-brightness areas – like the sky.

See also: Dynamic Rounding


Cyclic Redundancy Check, an advanced checksum technique used to recognize errors in digital data. It uses a check value calculated for a data stream by feeding it through a shifter with feedback terms ‘EXORed’ back in. It performs the same function as a checksum but is considerably harder to fool.

A CRC can detect errors but not repair them, unlike an ECC. A CRC is attached to almost any burst of data which might possibly be corrupted. On disks, any error detected by a CRC is corrected by an ECC. ITU-R BT.601 and 709 data is subjected to CRCs. If an error is found the data is concealed by repeating appropriate adjacent data. Ethernet packets also use CRCs.

See also: Asynchronous, Checksum, ECC, EXOR

DCT (compression)

Discrete Cosine Transform. As a basic operation of MPEG video compression it is widely used as the first stage of compression of digital video pictures. DCT operates on blocks (hence DCT blocks) of the picture (usually 8 x 8 pixels) resolving their content into frequencies, giving an amplitude for each frequency component. In itself DCT may not reduce the amount of data but it prepares it for following processes that will. Besides MPEG, JPEG, VC9, WM9 and DV compression all depend on DCT. The use of blocks can lead to blocks being visible on screen where data rates that are too low are used.

See also: DV, ETSI, JPEG, MPEG-2, MPEG-4, Wavelet


Tests to check the correct operation of hardware and software. As digital systems continue to become more complex, built-in automated testing becomes an essential part of the equipment for tests during both manufacture and operation. This involves some extra hardware and software to make the tests operate. Digital systems with such provisions can often be quickly assessed by a trained service engineer, so speeding repair.

Remote diagnostics can make use of an Internet connection to monitor and test a product at the customer’s site while working from a central service base. Thus expert attention can be used immediately on site.

Interdependent multiple systems, such as a video server and its clients, may require simultaneous diagnostics of all major equipment. Here, combining data links from a number of pieces of networked equipment, as with Quantel’s R-MON, effectively extends the Remote Diagnostics to larger and more complex situations.

See also: Diagnostics

Digital keying and chroma keying

Digital keying differs from analog chroma keying in that it can key uniquely from any one of the billion colors of component digital video. It is then possible to key from relatively subdued colors, rather than relying on highly saturated colors which can cause color-spill problems on the foreground.

A high quality digital chroma keyer examines each of the three components Y, B-Y, R-Y or R, G, B of the picture and generates a linear key for each. These are then combined into a linear key for the final keying operation. The use of three keys allows much greater subtlety of selection than is possible with a chrominance-only key.

See also: Chroma keying, Keying

Digital Leader and Digital Projection Verifier

Digital Leader and the Digital PROjection VErifier (DPROVE) are two products that are based on SMPTE RP 428-6-2009. The Digital Leader is aimed at digital movie post production and cinemas. In post it can be added as a leader and/or footer (end) of Digital Cinema Distribution Master (DCDM) ‘reels’ so allowing a quick quality check.

DPROVE is a set of Digital Cinema Packages (DCPs) that help checking projector performance and aligment as well as the sound’s synchronization with the pictures.

See also: DCI

Digital mixing

Digital mixing requires ‘scaling’ each of two digital signals and then adding them. A and B represent the two TV signals and K the positional coefficient or value at any point of the mix between them (i.e. equivalent to the position of the transition arm on a switcher desk). In a digital system, K will also be a number, assumed here as 10-bit resolution to provide a smooth mix or dissolve.

Mathematically this can be shown as:

A x K = (Mix)1
B x (1-K) = (Mix)2
Result = (Mix)1 + (Mix)2

Note that such math also applies to soft-edge keys and any transparency created between two images. As such it is a fundamental part of video processing and good quality results are essential.

When two 10-bit numbers are multiplied together, the result is a 20-bit number (see Binary). When mixing, it is important to add the two 20-bit numbers to obtain an accurate result. This result must then be truncated or rounded to 10 bits for transmission to other parts of the digital system.

Truncation by simply dropping the lower bits of the partial result (Mix)1 or (Mix)2, to 12 bits, or even 14 bits, will introduce inaccuracies. Hence it is important that all partial results, e.g. (Mix)1 and (Mix)2, maintain 20-bit resolution. The final rounding of the result to 10 bits can reveal visible 1-bit artifacts – but these can be avoided with careful rounding techniques such as Dynamic Rounding.

See also: Binary, Dynamic Rounding


A system which converts an analog input to a digital representation. Examples include analog to digital converters (ADCs) for television signals, touch tablets and mice. Some of these, mouse and touch tablet for example, are systems which take a spatial measurement and present it to a computer in a digital format.

See also: A/D, Into digits (Tutorial 1), GUI

Digitizing time

Time taken to record existing footage into a disk-based editing system. The name suggests the material is being played from an analog source, which is now rare. A better term is ‘loading’. Today the use of high-speed networking can enable background loading – eliminating digitizing time at the edit suite.

Digitizing time is often regarded as dead time, but it need not be. It can be reduced if some initial selection of footage has been made – for example by logging. Also, footage can be marked while loading and so be instantly available as a rough cut on completion, so speeding the final edit. The process is sometimes referred to as Triage, particularly where it is used to select and pre-edit clips from a live feed.

Discontinuous 2:3

It is common for the electronic editing process to be performed post telecine. When editing is performed to 2:3 there is a potential for disruptions in the 2:3 sequence. These can be 3 field sequences adjacent to other 3 field sequences, and 2 field sequences adjacent to other 2 field sequences. Also there are cases where we have single fields present that are not part of any sequence (Orphan fields). These disruptions caused by editing create a “broken 2:3 sequence”.


In digital television, analog original pictures are converted to digits: a continuous range of luminance and chrominance values is translated into a range of finite numbers. While some analog values will correspond exactly to numbers, inevitably others will fall in between. Given that there will always be some degree of noise in the original analog signal the numbers may dither by one Least Significant Bit (LSB) between the two nearest values. This has the advantage of providing a means by which the digital system can describe analog values between LSBs to give a very accurate digital rendition of the analog world.


If an image is produced by a computer, or is the result of digital processing, it may have virtually no noise and so the digital dither may not exist – which can lead to contouring effects. One approach to cure this is to use greater bit depth, eg 10 bits instead of 8, this will reduce the size of the problem but may not solve it. Another approach is offered with Dynamic Rounding. Invented by Quantel, this can intelligently add dither to pictures to give more accurate, better looking results.


Field dominance defines whether a field type 1 or type 2 represents the start of a new interlaced TV frame. Usually it is field 1 but there is no fixed rule. Dominance may go unnoticed until flash fields occur at edits made on existing cuts. Replay dominance set the opposite way to the recording can cause a juddery image display. Much equipment, including Quantel’s, allows the selection of field dominance and can handle either.

Drop-frame timecode

Alteration of timecode to match the 1000/1001 speed offset of NTSC transmissions and many newer HD and UHD video formats used in ‘NTSC’ countries – including the USA, Canada and Japan. 525-line NTSC at a nominal 30 f/s actually runs at 29.97 f/s, 720 and 1080-line HD as well as 2K and 4K UHD all include the 1000/1001 offset frequencies of nominal 24, 30 and 60Hz frame rate. So even 24 f/s movies run at 23.97 Hz.

With drop-frame timecode, the timecode is locked to the video and it needs to make up 1 in 1001 frames. It does this by counting two extra frames every minute while the video remains continuous. So 10:35:59:29 advances to 10:36:00:02. In addition, at every ten-minute point the jump is not done. This brings the timecode time almost exactly into step with the video.

Timecode that does not use drop-frame is then called non drop-frame time-code. Confusion arises when the wrong one is used!

See also: 1000/1001

Website: www.dropframetimecode.org

Dynamic Rounding

Dynamic Rounding is a mathematical technique devised by Quantel for truncating the binary word length (the number of 1s and 0s) of pixels to a specified number of bits. Rather than simply ignoring the lower bits, it uses their information to control, via a randomizer, the dither of the LSB of the truncated result. This effectively removes the artifacts, such as banding, that could otherwise be visible. Dynamic Rounding is non-cumulative on any number of passes and produces statistically correct results. Earlier attempts at a solution have involved increasing the number of bits (e.g. from 8 bits to 10 bits), making the size of LSBs smaller but not removing the inherent problem.

Some form of limiting the number of bits is required as there are many instances in digital systems where a number, representing a pixel value, uses more bits than the system normally accommodates. For example, a nominally 12-bit system might have a problem handling a 24-bit word. This has to be rectified in a way that will keep as much information as possible and not cause noticeable defects even after many processes. A common example arises in image processing which often involves multiplying the values of co-located pixels in two different images, as in digital mixing. Assuming the equipment is nominally 12-bit, the mixing produces a 24-bit result from two original 12-bit numbers. At some point this has to be truncated, or rounded, back to 12-bits, either to fit within the structure of the equipment or to be accepted by external interfaces. Simply dropping the lower bits can result in unwanted visible artifacts, especially when handling pure, noise-free, computer generated pictures.

16 bits to 8 bits

Dynamic Rounding

Dynamic Rounding is licensable from Quantel and is used in a growing number of digital products both from Quantel and other manufacturers.

See also: Digital mixing,


ErrorCheckand Correct. This system appends check data to a data packet in a communications channel or to a data block on a disk, which allows the receiving or reading system both to detect small errors in the data stream (caused by line noise or disk defects) and, provided they are not too long, to correct them.

See also: Checksum, CRC

Error detection, concealment and correction

No means of digital recording is perfect. Magnetic tape, disks and even perhaps SD (chips) suffer from a few marginal areas where recording and replay is difficult or even impossible. However the errors can be detected and some remedial action taken by concealment or correction. The former attempts to hide the problem by making it less noticeable whereas the latter actually corrects the error so that perfect data is output.

When the recorded data is an image, an error can simply be concealed by using data from previous or following TV lines, fields or frames. The result is not guaranteed to be identical to the original but the process is relatively simple and, as important, quick. If the stored information is from a database, a computer program or from special image processing, then 100% accuracy of data is essential. This can be ensured by recording data in a manner where any errors can be detected and the correct data calculated from other information recorded for this purpose. This is error correction.

A difference between computer systems and TV is that the latter is continuous and cannot wait for a late correction. Either the correct result must be ready in time or some other action taken, the show must go on, placing a very tight time constraint on any TV-rate error correction. In contrast, a computer can usually afford to wait a few milliseconds.

Digital VTRs monitor the error rate and provide warnings of excessive errors, which although not immediately visible, may build up during multiple tape passes.

Although error rates from disks are generally many times lower than those expected from digital videotape, they can still occur. To protect against this there is data redundancy and the replay of all data is checked. If an error is detected there is sufficient additional information stored to calculate and substitute the correct data. The total failure of a disk drive can be covered and the missing data re-generated and recorded onto a new replacement, making the system highly accurate and very secure.


ETSI compression

A compression technique, based on DCT. Unlike MPEG, which is asymmetrical having complex coders and simpler decoders and is designed for broadcast, this is symmetrical with the same processing power at the coder and decoder. It is designed for applications where there are only a few recipients, such as contribution links and feeds to cable head ends. ETSI compression is intra-frame, simpler than MPEG and imposes less delay in the signal path, typically 120 milliseconds against around a second, enabling interviews to be conducted over satellite links without unwarranted delays. Data rate is 34 Mb/s.


The mathematical operation of ‘EXclusive OR’ logic gate on a number of data bits. For example the EXOR of two bits is 1, only if one of them is 1 and the other 0. The EXOR is widely used in data recovery (see RAID). If the EXOR of a number of blocks of data is stored, when one of those blocks is lost, its contents can be deduced by EXORing the undamaged blocks with the stored EXOR.

See also: Error detection


1) A failure or disappointment.

2) A floating point calculation per second. Flops are a measure of processing power and today this runs into Gigaflops. For example, the powerful Cell processor used in PlayStation 3 is rated at somewhere between 218 and 250 Gigaflops.


The scattering of data over a (disk) store caused by many recording and deletion operations. Generally this will eventually result in store access becoming slow – a situation that is not acceptable for video recording or replay. The slowing is caused by the increased time needed to access randomly distributed data or free space. With such stores de-fragmentation routines re-arrange the data (by copying from one part of the disk to another) so that it is quickly accessible in the required order for replay. Clearly any change in replay, be it a transmission running order or the revision of an edit, could cause further fragmentation.


Stores capable of true random access, such as Quantel’s sQ server, are able to play frames in any order at video rate, and so never need de-fragmentation.

See also: Consolidation, FrameMagic

Frame rate

The number of whole pictures per unit time, usually frames per second, f/s or Hz. There is a wide range of frame rates used with modern media, typically from 23.976 to 60Hz. An even higher frame rate of 120Hz is included in ITU-R BT.2020 recommendation for UHD. This is because the large 4K and 8K video requires high frame rates to provide good smoothness of motion when looking at very big pictures.


Quantel term describing an advanced form of the management of video in a server. This covers much ground but basically offers the practical goals of guaranteeing realtime access to any frame for all video connections (simultaneous true random access) and avoiding the deletion of any material by one user that is partly or wholly used by another.

This is achieved by implementing a number of basic design criteria, including realtime random access to any frame, storing video material as a series of individual frames, rather than longer video files, as well as an on-board realtime database management system which, among other things, tracks who is using what material. FrameMagic is implemented in Quantel’s sQ servers.

See also: Delta Editing


The name, coined by Quantel, given to solid-state video storage, usually built with DRAMs. Technically it implies storage of one complete frame or picture, but the term is also used more generically to encompass the storage of a few lines to many frames. With large DRAM capacities available, framestores may hold many frames or a whole video clip, they are widely used to enhance equipment performance by providing instant access to video material.

See also: DRAM


The number of times and event occurs over a given period of time (usually one second). In most cases frequency relates to a regularly occurring event of a cyclic nature. Frequency is measured in Hertz , Hz, which is the SI unit defining cycles per second. It is named after the German physicist Heinrich Hertz who was not the founder of a car rental company we understand , A 440Hz tone describes the pitch of A4 audible tone. In electronic terms we often talk in terms of MHz, 10^6, and GHz, 10^9, cycles/second. For example, a specific frequency defines the clock rates in digital systems. Here are some of the more common ones used in TV:

PAL subcarrier: 4.43 MHz
NTSC subcarrier: 3.58 MHz
ITU-R BT.601 luminance sampling rate: 13.5 MHz (SD)
ITU-R BT.601 chrominance sampling rate: 6.75 MHz (for 4:2:2 SD sampling)
ITU-R BT.709 luminance sampling rate: 74.25 MHz (HD)
ITU-R BT.709 chrominance sampling rate: 37.125 MHz (for 4:2:2 HD sampling)
ITU-R BT.2020 luminance sampling rate: 297 MHz (4K UHD)
ITU-R BT.2020 chrominance sampling rate: 148.50 MHz (for 4:2:2 UHD sampling)

Although not appearing in any prominent headline, 2.25 MHz is significant as the greatest common divisor multiple of all these frequencies, meaning they are all related.

See also: 13.5 MHz


Gamma has several meanings. In the video world a television screen’s brightness is not necessarily linearly proportional to amplitude of the picture signal. This is approximated by a power law, the power being referred to as gamma. For a CRT (cathode ray tube) the gamma is generally calculated to be 2.6. This is compensated for in TV cameras by a gamma of 0.45 giving an overall gamma of 0.45 x 2.6 = 1.17 – adding overall contrast to help compensate for domestic viewing conditions. Today, most viewers have panel screens that use one of several different technologies. Usually there is a menu somewhere for selecting a gamma setting to suit your screen for the correct gamma and gamut.

In film, gamma describes the average slope of the D/Log E curve over its most linear region. For negative stocks this is approximately 0.6, for intermediate stocks this is 1.0 and for print stocks 3.0. This gives a system gamma of 0.6 x 1 x 3 = 1.8. This overall boost in contrast is much reduced due to flare and auditorium lighting conditions of cinemas.

With video now available on a wide range of devices there may be a need to re-visit the gamma settings. For example, a digital film master is versioned for mobile phones, and for home TV (where viewers have, LED, plasma and a few CRT displays) as well as for digital and film cinemas. This can be achieved by applying suitable LUTs for each version.

See also: Internegative, Interpositive

Website: www.myperfectpicture.tv


In image media this usually refers to the available range of colors from a display device, as in Color Gamut. This varies according to the color space used; YCrCb, Rec709 (HDTV), Rec2020 (UHD), DCI-P3 (digital cinema) and CMYK (paper print colors) all have different color gamuts.

See also: Illegal colors


A device connecting two computer networks. For example, a gateway can connect a local area network (LAN) to a storage area network (SAN). This way a PC connected on an Ethernet LAN may have access to files stored on the SAN even though the PC is not SAN aware.

Generation loss

The signal degradation caused by successive re-recordings (copying). Freshly recorded material is first generation; one re-recording, or copy, makes the second generation, etc. This was of major concern in analog linear editing but much less so in a digital suite. Non-compressed component DVTRs should provide at least twenty tape generations before any adverse artifacts become noticeable, most likely due to dropouts on the tape. Even better multi-generation results are possible with disk-based systems. These can re-record millions of times without causing dropouts or errors. This means that the number of generations are effectively limitless, which is very useful when building multi-layer video, or movie, effects.

Other possible recording media candidates include solid state memories such as DRAM, which is very reliable over a huge number of read/write cycles; provided the power stays on. Static RAM, such as SD cards, is not suitable as it is not as fast as DRAM and is known to degrade over many read/write cycles.

Besides the limitations of recording, the action of processors such as decoders and coders will make a significant contribution to generation loss. The decode/recode cycles of NTSC and PAL are well known for their limitations. Caution is also needed for digital video compression (coding) systems, including MPEG, DV and JPEG, not to mention the vast number of proprietary variants, as well as the color space conversions that typically occur between computers handling RGB and video equipment using Y,Cr,Cb.

See also: Concatenation, Error detection concealment and correction


General Purpose Interface. This is a simple form of control interface typically used for cueing equipment, usually by a contact closure. It is simple, can be frame accurate and therefore can easily be applied over a wide range of video equipment.


Graphics Processing Unit. A chip or digital circuit designed specifically for processing graphics and generally providing the main processing power of a computer graphics card. Having much more graphics power and speed than central processor unit (CPU) chips, GPUs can take over many complex 2D and 3D image processing tasks from the CPU.

In the wider computer market, GPUs have transformed computer games, simulation and a whole host of complex digital displays. However, GPUs are not as versatile as CPUs and some graphics tasks may still need to go through CPUs or other specific processors.

In the media business, GPUs are used in a wide range of image processing equipment – such as picture color correctors and graders as well as the high-end television graphics equipment including that for live ‘3D’ graphics and virtual sets.


Term describing limits of accuracy or resolution. For example, in editing the granularity of uncompressed component video is one frame; it can be cut on any frame boundary. The granularity of long GOP MPEG-2 is about half a second; about 12 or 15 frames for 25 or 30 f/s television. In a digital imaging system the granularity of brightness is the minimum change per sample, corresponding to the effect of one LSB change.

HD ready

This describes a television that can display the recognized 720 and 1080-line formats but does not include the tuner or decoder needed to receive the signals. Typically the screen will not have the full 1080-line resolution so such video will be up-res’d.


A numbering system, often referred to as ‘Hex’, that works to base 16 (instead of our usual base 10) and is particularly useful as a shorthand method for describing binary numbers. Decimal 0-9 are the same as Hex, then 10 is A, 11 is B, up to 15 which is F.


See also: Binary


High Frame Rate – a frame rate higher than normal. For instance, movies (films) are normally shot at 24 f/s but some have been shot at 48 f/s – HFR. Some audiences say they do not like it as it’s too real and does not look like film.

It has been observed that when viewing UHD, motion judder is often very apparent and so a higher frame rate (say 48 f/s) is recommended by some. When shooting fast-action sports, such as football, then the UHD result would look better using, say, 50 or 60 f/s. In fact the UHD standard Rec 2020 includes frame rates up to 120 f/s.

Huffman coding

This compresses data by assigning short codes to frequently occurring long sequences and longer ones to those that are less frequent. Assignments are held in a Huffman Table. Huffman coding is lossless and is used in video compression systems where it can contribute up to a 2:1 reduction in data.

See also: JPEG

Illegal colors

Colors that lie outside the limits, or gamut, of a particular defined color space. These can be generated when transferring images from one color space to another, as they all have different boundaries, or as the result of color processing. For example, removing the luminance from a high intensity blue or adding luminance to a strong yellow in a paint system may well send a subsequent video signal too high or low, producing at least inferior results and maybe causing technical problems. Out-of-gamut detectors can be used to warn of possible problems and correction is also available. Some broadcasters reject material with illegal colors.

Interlace Factor

The reduction in vertical definition during vertical image movement due to interlaced (rather than progressive) scans. When interlace is used in broadcast television it means that a field of all the odd lines (1, 3, 5, etc) is sent followed by the even (2, 4, 6, etc.) lines that fit between the odd ones. Experimentally the interlace factor is found to be about 30%. Note that, when scanning film frame-per-frame (i.e. 24 or 25fps, not 3:2 pull-down to 60fps), or a succession of electronic frames each representing a single snapshot in time, there is no vertical movement between fields and so in these cases the Interlace Factor has no effect.

See also: 24PsF

Interlace (scan)

Method of scanning lines down a screen (vertical refresh). It is still used in most of today’s television broadcasts but was originally designed to suit the needs of CRT displays and analog broadcasts. Interlace is indicated in television scan formats by an ‘I’ e.g. 1080I, etc. (though the use of ‘i’ is common). Each displayed picture comprises two interlaced fields: field two fills in between the lines of field one. One field displays odd lines, then the other shows even lines. For analog systems, this is the reason for having odd numbers of lines in a TV frame eg 525 and 625, so that each of the two fields contain a half-line, causing the constant vertical scan to place the lines of one field between those of the other.

The technique greatly improves the portrayal of motion and reduces picture flicker without having to increase the picture rate, and therefore the bandwidth or data rate. Disadvantages are that it reduces vertical definition of moving images by about 30% (see Interlace Factor) of the progressive scan definition and tends to cause some horizontal picture detail to ‘dither’ – causing a constant liveliness even in still pictures.

Interlaced video requires extra care for processing, such as in DVE picture manipulation of size, rotation, etc, as any movement between fields has to be detected if frame-based processing which can produce higher-quality results, is used. Also frame freezes and slow motion need ‘de-interlace’ processing.

There is continuing debate about the use of interlaced and progressive scans for digital television formats. This has intensified now that the increasingly popular panel displays all use progressive scans. Interestingly the latest television standard for ITU-R BT.2020, for 4K and 8K UHD only includes progressive scans.

See also: Interlace Factor, Progressive


Integrated Receiver Decoder. A device that has both a demodulator and a decoder (e.g. for MPEG-2) built in. This could be in a digital television set or a set-top box.

See also: IDTV


Integrated Server Architecture is a Quantel term for the technology used in its sQ servers to manage the contents of several separate servers simultaneously. ISA operates at two levels: one locks browse and full quality material together under a single ISA database. The other level allows all material held on several sQ servers to be handled as if on a single server, effectively as if on a single database. This facilitates system scaling of users and storage by adding servers and makes possible the ‘ZoneMagic’ operation where two separated servers are kept in step.

See also: NAS, SAN


A form of data transfer that carries timing information with the data. Data is specified to arrive over a time window, but not at any specific rate within that time. ATM, IEEE 1394 and Fibre Channel can provide isochronous operation where links can be booked to provide specified transfer performance. For example, 60 TV fields can be specified for every second but their arrival may not be evenly spread through the period. As this is a guaranteed transfer it can be used for ‘live’ video but is relatively expensive on resources.

See: ATM, Asynchronous, Fibre Channel, IEEE 1394, Synchronous


A general purpose programming language developed by Sun Microsystems and best known for its widespread use in animations on the World Wide Web. Unlike other software, programs written in Java can run on any platform type, so long as they have a Java Virtual Machine available.

Website: java.sun.com


Just a bunch of disks. This could be a collection of disk drives connected on a single data bus such as SATA, Fibre Channel or SCSI. JBODs are cheap and can offer large volumes of storage that may be shared among their users. As there are no intelligent controllers, items such as data speed and protection may well be compromised.

See also: SAN

Latency (of data)

The delay between requesting and accessing data. For disk drives it refers to the delay due to disk rotation only, even though this is only one of several factors that determines time to access data from disks. The faster a disk spins the sooner it will be at the position where the required data is under the replay head. As disk diameters have decreased so rotational (spindle) speeds have tended to increase but there is still much variation. Modern 2.5-inch drives typically have spindle speeds of between 7,200 and 15,000 RPM, so one revolution is completed in 8 to 4 ms respectively. This is represented in the disk specification as average latency of 4 or 2 ms.

For solid-state ‘drives’ (SSD) the latency is much less resulting in faster access to data, particularly if the data is scattered around the store; here, SSD can be over 100 times faster than HDD, but just a few times faster for large unscattered data files.


Liquid Crystal On Silicon. An imaging chip technology that has been likened to a cross between LCD and DLP. Like LCDs this uses one liquid crystal per pixel to control the light, but whereas LCD is transmissive, the light travels through the crystals, LCOS appears as a reflective chip, like DLP. LCOS is the basis for many imagers, JVC’s implementation is D-ILA and Sony’s SXRG appear to use some similar ideas – though many refinements are used.


Least Significant Bit. In a binary number, this is the final digit – 0 or 1 – and is the least significant because changing it has the smallest effect on the whole number represented compared with changing any of the other digits in the binary number. However, even with their relative insignificance, LSBs can have a significant impact if not cared for properly, particularly when multiplying two binary numbers together – which is why Quantel invented Dynamic Rounding.

See also: Dynamic Rounding, Binary, Dither, Truncation, Digital Mixing, MSB


A component of video: the black and white or brightness element, of an image. It is often represented as Y, so the Y in Y,B-Y,R-Y, YUV, YIQ and Y,Cr,Cb is the luminance information of the signal.

In a color TV system the luminance signal is usually derived from the RGB signals originating from cameras, by a matrix or summation of approximately:

Y = 0.3R + 0.6G + 0.1B (based on ITU-R BT.601)

There are other similar equations from different TV standards. The precise values depend on the color primaries of the display standard used.

See also: RGB, Y (B-Y) (R-Y), Y,Cr,Cb, YUV


LUT is the shortened description of the term ‘Look-Up-Table’. In its simplest form it consists of a finite number of  positive integer input values which map to new output values. Typically a LUT will take 2n (64, 128, 256…) input values from 0 to 2n – 1 and map the input to the new output. Although they can be any size required. For example, a small LUT of size 16 might look like this when describing a math function where the output = input2 :



The example shows positive integer address (input) producing an integer Output. This need not be the case; the output could be floating point or any number form needed.

Although the sample LUT above is a mathematical model this does not have to be the case as the input can refer to any value the user desires so LUTs are very convenient where there may be no distinct or a very complex connection between the input values and the output.

LUTs are very common in hardware based systems as they can be used to simplify a mathematical process to map Input address to Output very efficiently. With today’s high performance GPU math processes the developer needs to consider whether the LUT is as efficient as just doing the math on each value.

Inputs to LUTs are typically positive integers. It is possible to engineer a more complex form of LUT which takes a floating point input. Here the LUT is used to look up integer values above and below the Input value and the final result is interpolated from there. For example:

Input value = 3.4

For LUT Input value 3 the Output = 27.5 (these values come from some other complex process)

For LUT Input value 4 the Output  = 29.5

The final output will be 27.5 + (29.5 – 27.5) * 0.4 = 28.3.

Thus a complicated math process can be approximated by a LUT and a simple interpolation.

This process can be expanded further into multi-dimensional LUTs where the Output is a result of multiple Input values.

In image processing systems LUTs can be used for gamma correction, color space conversion (rgb to XYZ, yuv to rgb) and color correction for example.

See also: Color cube


The master of an edited program holds the material that is used as the source for making all deliverables and other versions (for language, subtitles etc.). Making a good quality master sufficient to fulfill quality needs of all target customers the standards should ensure that the deliverables are good.

For international distribution the use of 1080 x 1920 24P (the same frame rate as movies) is often regarded as the best to ensure good quality deliverables for HD and SD requirements. If it is a high-cost production aimed at a world market a 4K UHD may be required. Also, 24P can be a good frame rate for drama, but many other genres, including spot, would require higher or 50 or 60 I or P.

Supplying the best to all media platforms now often requires more than the traditional master can provide. A better form of master may be an uncommitted one, where all the original source material and all the tools used and their settings, are available so that any aspect of editing and finishing can be revisited to make the right deliverables for everyone.

See also: Deliverables


Software, not hardware. This exists above an operating system to provide a middle layer offering APIs for applications programmers but it is not an application itself. An example is Multimedia Home Platform (MHP) which is widely used in set-top boxes.


Short for modulator/demodulator, it is a two-way communications interface working between a communications channel, such as a DSL line, and a machine such as a computer. That is how billions of people access the internet. Television itself is distributed live via a modulator in the transmission chain, and a demodulator at the receiving end.

The efficiency of modern modems over analog is worth noting. In analog days it generally used PAL or NTSC ‘coding’, that compressed three channels R, G and B into one, which was transmitted in one TV channel. Today there are very efficient compression systems such as MPEG-2, MPEG-4, H.264 and H.265 used in DVB, ATSC or other digital systems. The combination of the latest video compression, H.265, and the efficiency of DVB-T2 modulation will allow up to 32 SDTV channels, or 4 HD channels, or one 4K UHD channel, to be delivered in the space (bandwidth) that was occupied by one analog PAL 625-line channel.


Refers to a distortion that appears as patterning seen on images where two similar fine patterns overlap, for example two fine gratings can appear to produce diagonal lines. The effect can appear even when one of the patterns is normally invisible, such as the sampling frequency of the image. In a good image system this should be avoided by use of filters but, for instance, the fine detail of a grid pattern may suddenly collapse and appear as curves or diagonal lines as the camera zooms in and the pattern detail nears the digital sampling frequency.

The occurrence of Moiré patterning changes with the picture scanning system. The higher the picture’s resolution the finer the pattern that will cause Moiré. For example, a quite small dog-tooth patten worn by a news presenter could cause it in SD, but with HD it has to be twice as fine.

Moore’s Law

A prediction for the rate of development of modern electronics. This has been expressed in a number of ways but in general states that the density of information storable in silicon roughly doubles every year. Or, the performance of silicon will double every eighteen months, with proportional decreases in cost. For more than two decades this prediction has held true. Moore’s Law initially talked about silicon but it could be applied to other aspects such as disk drive capacity that doubles every two years and has held true, or been exceeded, since 1980, and still continues unabated.

See: Disk drives

Motion Adaptive Processing

Motion adaptive processing is responsive to the output of a motion detector.  The motion detector may work on a global or a local basis in the picture, and may be binary or may measure on a continuous scale the amount of motion, or the confidence that an object or region is moving.  Motion adaptive processing controls a mix between processing that is optimized for static detail and processing that is optimized for moving areas.

Motion Compensated Processing

Motion compensated processing is responsive to the output of a motion estimator.  A motion estimator usually works on a local basis in the picture and measures not only the existence of motion but also its speed and direction.  Motion compensated processing typically controls spatiotemporal filters that track the motion of each part of the picture.