Contents

Plug: See also my list of links related to video and television.

What this document covers and what it doesn't

There seems to be a lot of confusion about the huge number of different video recording formats and their specifications. Information is widely scattered and hard to find. This document will mostly cover professional and semi-pro formats, both those in use now and those being under development. If you want to know about "consumer" formats, check out this.

For the time being, there isn't much information on analogue formats, but I will be adding more stuff as I come across it. Unless noted, all information applies to PAL 625/50 versions. I'm omitting SECAM on purpose, because it's rarely used for studio work anyway.

Many of the figures have been extracted from manufacturers' brochures and do not necessarily represent the best a given recording format can offer. Nevertheless, they should be pretty close.

And to keep things in perspective, let me remind you that 16 mm film beats the living daylights out of most formats depicted here, especially when it comes to dynamic range.

Notes on specifications and nomenclature

Broadcast quality defined - not!

The rest of this document, as well as many other texts handling video production, will frequently refer to "broadcast quality" without really defining what it is. Such a definition doesn't really exist -- many people think of it as describing the minimum quality for a program that is to be broadcast. There is no technical standard for this, and in fact, almost any crap can be restored to good enough quality with time base correctors and other digital processing.

From the technical point of view, broadcast quality can be thought of as a video signal fulfilling the timing and signal level tolerances placed by the relevant international standards. This doesn't say anything about the bandwidth (resolution) or signal-to-noise ratio of the actual picture -- that remains to be judged from the highly untechnical point of view: considering the contents of the program, is the picture watchable enough? Now that there are plenty of good recording formats available, broadcast quality recording usually means something that isn't noticeably worse than direct composite video from a good camera.

Composite vs. component

In this context, component video means colour video represented by three separate signals (luminance and two colour difference signals). This is commonly referred to as Y/Cr/Cb colour space. RGB is almost never used in recording, because it requires 3/2 times the bandwidth of Y/Cr/Cb representation in order to achieve similar subjective quality. Some video equipment may still have RGB inputs or outputs, as converting between these two component representations is easy; see the equations at Rick Davis' digital video page (that's for NTSC, by the way -- PAL uses slightly different coefficients). Charles Poynton's excellent FAQs on colour and gamma should answer the rest of your questions.

S-Video (or Y/C) does not count as component video, as most externally composite formats use separate chrominance and luminance signals in recording anyway. Additionally, the chrominance signal of S-video is already modulated in either PAL or NTSC to ride on a subcarrier and as such is limited in bandwidth. Component video does not have this limitation.

Bandwidth and lines

Both of these figures refer to the horizontal resolution. "Lines" refers to the number of vertical alternating black and white lines that can be stuffed in the picture and still be perceived as separate lines and not a gray mass. It's clear that this is not a very scientific definition of resolution when the signal is in the continuos domain, like it is with all analogue formats. Therefore, the bandwidth is better expressed as the real electrical signal bandwidth fitting inside some dB limits, which can be easily measured.

As you are bound to come across bandwidth specifications in both lines and frequencies, here's how to convert from lines to bandwidth: 

 
           4/3 * resolution in lines
    BW = -----------------------------  / 2.
         length of active picture line

The divisor 2 comes from the fact that you need two lines (black and white) to represent a single sine wave cycle. To make things complicated, even the horizontal line resolution is usually expressed as the number of lines that could be reproduced vertically, if the horizontal and vertical resolutions were equal. Therefore the equation contains the multiplier 4/3, which comes from the aspect ratio of television screen. I doubt that all advertisers and salesmen know this.

For 625/50 PAL, the length of the active (visible) part of the picture line is 52 us, so for example the 240 line horizontal luminance resolution of VHS becomes 

 
    4/3 * 240
    --------- / 2 = 3 MHz
      52e-6

worth of bandwidth. Of course, we don't know the dB limits for this figure, because nobody told us how gray those 240 "black and white" lines got. For this reason, respectable +/- 1 dB bandwidth specifications may look worse than the same picture quality expressed in lines (which is why lines are preferred in advertisements). Reversing the equation to get from bandwidth to lines is left as an exercise to the reader.

In the specs for component formats, the expressed colour bandwidth applies to both of the colour components separately.

Note: This simplified discussion omits some important points, but will suffice for the usual 4:3 aspect ratio TV systems. If you want the dirty details, consult a good text book like [2]. Furthermore, this conversion method isn't valid for chrominance bandwidth and resolution (I'm still trying to figure out exactly why, but I suppose it's because the colour bandwidth in composite video is limited to some 1.5 MHz in any case).

But what about vertical resolution?

Unlike the horizontal direction, the vertical direction of television picture is discrete and not continuous. Still, on the picture tube the lines overlap and are not normally perceived as separate. For this reason (and others, like the Kell factor, but I've digressed enough already from the original point of this document), the vertical resolution of TV is not such a big deal. If it were, people would have abandoned 525 line NTSC long time ago. As all of the analogue formats record picture lines totally independently of each other (with some exceptions, see below), there is no need to state the vertical resolution. It is always the same as in the video format itself -- 575 visible lines in PAL, 485 in NTSC.

The same story applies to the digital formats which are represented here. None of them throw away every other line, like CD-i for example does. Even the formats using compression maintain the entire vertical resolution, although all the compression methods are two-dimensional and operate on the image as a whole and not on separate lines.

One can argue that that in PAL, the principle used for colour phase error cancellation reduces the vertical chroma resolution by half. But, this doesn't happen until in the receiver - the video recorders and the PAL video itself still carry the full vertical colour resolution. The exceptions here are VHS and S-VHS, where the method used for track-to-track chrominance crosstalk reduction really cuts the vertical chroma resolution to half on the tape. The chrominance also moves down in the picture. This is hardly notable in the first generation (cf. PAL), but unfortunately degenerates during each copy, so second and third generation VHS copies get barely acceptable and poor, respectively. The results can be seen as those famous "carved-with-an-axe" images.

Analogue recording formats

 
Format        M-II          EBU-C         Betacam SP    U-Matic SP   S-VHS
------------------------------------------------------------------------------
Type          Component     Composite     Component     Composite    Composite
              (Y,Cr,Cb)
Bandwidth
  Luma        5.0 MHz       5.5 MHz       4.5 MHz       ~ 4 MHz      ~ 5 MHz
  Chroma      1.8 MHz       1.5 MHz       1.5 MHz                    ~ 600 kHz
S/N
  Luma        > 47 dB       43 dB         > 51 dB       > 46 dB      45 dB
  Chroma      > 50 dB       43 dB         > 53 dB       > 48 dB
 
Audio         2 x FM,       3 x linear    2 x linear    2 x linear,
              2 x linear  (2 + time code) 2 x AFM, TC   time code
  S/N         > 85 dB                     > 72 dB       > 52 dB      > 90 dB
              (FM)                                                   (HiFi)
 
Tape          1/2" cass.    1" open reel  1/2" cass     3/4" cass    1/2" cass
Tape speed    6.6 cm/s      24.4 cm/s     10.15 cm/s    9.53 cm/s    2.399 cm/s
                                                                     (SP)

Uncompressed digital recording formats

 
Format             D-1       D-2       D-3       D-5       D-5        D-6
-------------------------------------------------------------------------------
Type               Component Composite Composite Component Component Component,
                                                                     HDTV
Sampling freq.                         4 Fsc =   Y: 13.5M  Y: 18M
                                       17.7 MHz  C: 6.75M  C: 6.75M
 
Quantization       4:2:2,    8 bits    8 bits    CCIR601,  4:2:2,
                   8 bits                        10 bits   8 bits
 
Bandwidth +-0.5dB                      
  Luma                                 6.0 MHz   Y: 5.75M  Y: 7.67M
  Chroma                                         C: 2.75M  C: 3.67M
 
S/N                                    54 dB     > 60 dB   > 56 dB
 
Data rate                                        270 Mbps
 
Audio              4 x 48k,            4x48k @        4x48k @
                   digital             16 bits        20 bits +
                                                      1 analog
Tape/speed         3/4"      3/4"      1/2",          1/2",
                                       8.4cm/s        16cm/s

Digital formats employing compression

With compressed digital formats, bandwidth doesn't make as much sense as without compression or with analogue video. Quantization (sampling) type and the compression ratio are more interesting; most of these formats use quantization according to CCIR Rec. 601, [1], which is widely considered as the specification for "broadcast quality" when it comes to digital video. The most common CCIR 601 subformat is 4:2:2 quantization with 720 active samples per line. 4:2:2 means Y/Cr/Cb (YUV) format video, where the sampling rate for both of the colour difference signals is half the sampling frequency of the luminance. 

 
Format           Quantization    Compression       S/N       Audio
----------------------------------------------------------------------------
Digital Betacam  4:2:2,          per field DCT,              4x20bit +
                 8/10 bit        2.3:1                       cue + TC
 
Digital S        4:2:2,          intraframe DCT,             4x16 bit
                 8 bit           3.3:1 @ 50 Mbps             48 kHz PCM
 
Ampex DCT        4:2:2,          DCT, 2:1
                 8 bit
 
DV/DVCPRO        4:2:0/4:1:1     5:1 @ 25 Mbps,    54 dB     2x16 bit/
                 (see below)     per field/frame             4x12 bit PCM
 
DVCPRO 50        4:2:2           3.3:1 @ 50 Mbps             4 digital + cue
 
Betacam SX       4:2:2           10:1 MPEG-2,                4x16 bit
                                 main level                  48 kHz
----------------------------------------------------------------------------

Some background on the formats

Since many of these recording formats, especially the digital ones, are almost unknown outside the broadcasting companies wealthy enough to use them, I've decided to include some background information on them too.

Betacam SP

Developed by Sony, perhaps the most popular component format for both field acquisition and post production today. Betacam uses cassettes and transports similar to the old Betamax home video format, but the similarities end there. Tape speed is six times higher, and luminance and chrominance are recorded on two separate tracks. The two colour difference signals are compressed in time by two and recorded sequentially on a single track.

M-II

M-II (read em-two) was developed by Matsushita for Japan's national broadcasting company NHK. Today M-II is one of the most popular broadcast quality component formats with quality similar to Betacam SP. Large users of M-II include NHK, of course, and NBC. Recording technique is similar to Betacam SP but uses some enhancements which compensate for the lower tape speed.

EBU C format

These machines use 1" tape in open reels. The main advantages are very fast transports and low recording density, which makes the format rather immune to drop-outs. Tape costs are high. The units can record single frames, which makes them popular in computer animation. Some units with vacuum capstans can operate from stop to nominal speed within one video field. The tape makes almost a full circle around the picture drum, and a single head is able to record and playback the entire video signal (short of a few lines right after vertical sync). Most other helical scan formats have at least two picture heads, which alternate between fields.

Note that in C format, the entire composite video signal is recorded and played back as is without splitting it to Y/C, like most composite recorders do, or limiting the bandwidth in any way. Numerous manufacturers include Sony, Ampex and BTS.

EBU B format

Similar to C format, but uses segmented helical scan. The diameter of the picture drum is small, and a single video field is recorded in 6 separate tracks. Manufactured by Bosch. B format doesn't allow for many special modes - play, FF and REW are just about it.

U-Matic

Another format by Sony. Has three different versions (LB, HB and SP), which differ by the subcarrier frequencies used for luminance and chrominance recording. U-Matic LB (Low Band) has been around from the early 70s and is one of the oldest cassette video formats. HB (High Band) has increased chroma subcarrier frequency, which improves colour resolution. In the SP variant, both chroma and luma subcarrier frequencies have been increased.

U-Matic SP (in common lingo "3/4" after the tape width in inches) is still a popular production format for those not wealthy enough to use Beta SP or similar. Although U-Matic doesn't appear much better than Super VHS on paper, the higher colour resolution and much better signal-to-noise ratio make the picture subjectively far more enjoyable. The U-Matic tape transport is also much faster in changing modes, which makes editing less frustrating.

LB and HB U-Matic tapes are often used for archiving because of the relatively low tape costs and low recording density, which makes the tapes robust against aging.

D series digital formats

D-1 was the first practical digital format, introduced by Sony in 1986. Although still considered a quality reference, D-1 is expensive to buy and use and has been mostly superseded by the more cost effective later formats.

D-2 was developed by Ampex around the same time as D-1 was introduced and is meant to be a fully transparent storage for composite video, useful for composing "spot tapes" for programmes such as news.

D-3 and D-5 have both been developed by Matsushita. D-5 units can use two different sample rate / resolution combinations and are generally capable of playing back D-3 tapes. While D-5 is still a studio format, D-3 camcorders are available from Panasonic.

A high definition version of D-5 (D-5 HD) has been introduced by Panasonic. It uses 4:1 compression in order to accomodate the 1.2 Gbps HDTV data rate. It can work with both 1080 line interlaced or 720 line progressive (American) HDTV formats.

D-6 is a digital HDTV recording format by Toshiba/BTS. Stores 600 GB worth of data on a physically huge 64 minute cassette. I was told that this format is more or less dead and the remaining tape transports were bought out by Toshiba - but in IBC'97, I saw a D-6 recorder in action, demonstrated as a joint venture between Toshiba and some other Japanese manufacturer, whose name escapes me. The picture quality was truly impressive.

DVCPRO should become D-7 later. Also, Digital-S will be designated as D-9 [unconfirmed].

As a curiosity, D-4 doesn't exist, as number 4 is a major taboo in Asian cultures (pronounced the same as "death" in Japanese). Rumours go that this even delayed the standardization of D-3 and D-5.

Digital Betacam

Digital successor to the venerable Betacam SP format. Introduced by Sony in 1993, uses physically similar half-inch cassettes. Camcorders with 40-minute capacity are available, making Digital Betacam the first component digital ENG (electronic news gathering) format. Digital Betacam units play back, but do not record analogue Beta SP tapes.

The 2:1 compression is based on DCT (discrete cosine transform), like most modern video compression techniques. Each field is compressed separately.

Ampex DCT

So far, fully proprietary format by Ampex (1992). Name is an acronym for "Digital Component Technology" and not for what you'd expect. According to Ampex, the main advantage of DCT over other digital formats is high tolerance for data errors, in the order of one uncorrected error per hour.

The units can switch between 525/60 and 625/50 operation, and allow a maximum recording time of more than three hours. These qualities give the DCT format a central role in Ampex' electronic film mastering scheme.

DV/DVCam/DVCPRO

DV (formerly DVC) is a new format being backed by manufacturers such as Sony, Philips, Thomson, Hitachi, Matsushita (Panasonic) and others. It was the first digital recording format in the reach of consumer markets. DV uses 5:1 compression based on DCT. Depending on the image contents, the encoder adaptively decides whether to compress picture fields separately or combine two fields into a single compression block. As such, DV coding can be thought of as something half-way between Motion JPEG and MPEG.

As a curiosity, the consumer version (DV) sports one of the densest recording techniques based on magnetic tape media - more than 0.4 megabits per square millimeter. Imagine the data from your 3.5" HD floppy recorded on a single-sided 5x6 mm piece of tape. New equipment from Sony will push this even further with the deployment of LP mode, which will reduce the track width to 6.67 um and multiply the recording time by 1.5. Video specifications will remain the same.

DVCPRO is a professional variant of the DV by Panasonic. The only major difference is doubled tape speed, which is needed for better drop-out tolerance and general recording robustness. It is also capable of 4x normal speed playback. This doesn't mean your run-of-the-mill FF with picture, but accelerated transfer of all of the information into for example a non-linear editing system. DVCAM on the other hand is Sony's variation of the theme, sitting somewhere between DV and DVCPRO. Tape speed and track width have been increased, but not as much as for DVCPRO. Furthermore, it uses the same metal evaporated tape as DV, while DVCPRO uses metal particle tape. What exactly Sony expects from this format is quite puzzling, given that they already have two other digital ENG formats - Betacam SX and Digital Betacam.

These similar, but still different digital formats have made some people to fear for another War of the Formats, the players this time being Panasonic and Sony. But because the formats mainly differ in the way they store the data on the tape, the data itself being the same, the situation isn't quite that sad. Actually Panasonic has announced that their upcoming DVCPRO gear will be able to play back DVCAM recordings; according to Panasonic, this will only require reprogramming the capstan servo systems to accomodate for the slightly different track width (15 um in DVCAM vs. 18 um in DVCPRO) and therefore different tape speed.

As for the picture quality, all these variants are nearly broadcast quality, DV being available at nearly consumer prices. For newsgathering and other similar uses, the quality is certainly enough, especially considering that typical postproduction will be done digitally, which will not degrade the quality any further. Compression is mild enough to keep artifacts away in all but problem scenes. The 4:1:1/4:2:0 quantization will be visible if you try something like chroma keying, however. As stated before, the picture quality in all three formats is identical - the targeted market segments are differentiated with the features of the equipment and things like available camera quality.

Recently, DVCPRO has been accepted to be standardized as D-7 by SMPTE. Panasonic has also introduced a 4:2:2 version: DVCPRO-50, which is intended to directly rival Digital S in the professional market.

Chroma subsampling methods used in the DV variants

Above, both 4:1:1 and 4:2:0 sampling methods have been mentioned. The former means that both of the chroma difference signals (Cr and Cb) are sampled at one quarter of the luminance sampling rate. The latter, on the other hand, uses half the luma sample rate for the colour differences, but the two differences are only sampled on alternating lines (resembling SECAM). Both give the same total data rate, but the 4:2:0 sampling gives a better apparent colour resolution for PAL.

In the NTSC world, all three DV variants use 4:1:1 sampling. In the PAL versions, DV and DVCAM revert to 4:2:0, whereas DVCPRO uses 4:1:1 globally. Here's a catch - if you dub between PAL DVCPRO and one of the other formats, you'll end up with the worst from both worlds: a 4:1:0 sampled image, where both vertical and horizontal colour resolutions are only half of the conventional broadcast quality 4:2:2 sampling.

A good source for further practical information on DV is DV Central.

Digital8

A recently introduced consumer format from Sony. Digital8 records for all practical purposes the same signal as DV, but uses cheaper Hi8 tapes and can play back old analogue 8mm/Hi8 tapes.

D-VHS

New format by JVC. This is a digital "bit bucket" format which is intended to store future digital broadcasts directly in its compressed format. Particularly, the machines will not have conventional video inputs and outputs at all - they work through a set-top decoder, just like direct digital broadcasts off the air. The machines will all have IEEE 1394 digital interfaces, like some DV format equipment do.

Longest tape should be able to store 44.4 GB of data. Recording times depend on the mode used, ranging from 3.5 hours of HDTV to 49 hours of 2 Mbps video in LP mode. Standard mode will record seven hours of video at 19.14 Mbps.

W-VHS

W-VHS was introduced by JVC in the 1994 NAB show. It is a analogue high definition format using a cassette physically identical with VHS.

Normally records 1125 lines of analog RGB video, but can switch to an NTSC mode, in which it can either record twice as much high quality NTSC video, or even two different video signals. The latter function is used for field sequential 3D video.

Digital S

Digital S is a digital format downward compatible with S-VHS. Cassette is basically a W-VHS high coercivity tape in a dust-proof version of the VHS cassette case. Digital S rivals the much more expensive Digital Betacam in terms of picture quality because of the mild compression and 4:2:2 quantizing.

JVC's Digital S editing deck sports a rarely seen feature, video pre-read head, which allows the old video recording to be played back while recording new signal just after that. This makes A/B roll edits possible with just two decks.

JVC also has a web site dedicated to Digital-S.

Betacam SX

Hot from the ovens at Sony, Betacam SX is a digital format using a 4:2:2 coded variant of MPEG. Like DVCPRO, it's capable of accelerated playback/recording and is part of Sony's new all-digital production concept, targeted especially for ENG and newsroom use. Some units are 4:3 / 16:9 switchable.

Despite primarily being a tape format, some decks are actually disk/tape hybrids and provide rudimentary stand alone non-linear editing capabilities. "Briefcase" field editors resembling laptop computers are available, similar to what Panasonic has for its DVCPRO format.

More hype is available on Sony's Betacam SX site.

Sony HDD-1000

This is a full bandwidth 1.2 Gbps digital HDTV recorder based on 1" open reel C format. Prototyped as early as in 1987, the format records 1125 line HDTV at either 59.94 or 60 Hz field rate. The steep price tag (~$350000 for the units, and 63 minutes of tape will set you back $1300!) makes this format somewhat of a curiosity, but it is both available and in use.

References

[1] CCIR (ITU-R) Recommendation 601-2, Encoding parameters of digital television for studios, Recommendations of the CCIR, 1990, Volume XI - Part 1. Geneva 1990

[2] Ru van Wezel, Video Handbook, 2nd edition. William Heinemann Ltd, London 1987

[3] Pentti O A Haikonen, Videotekniikka

[4] TV Technology magazine, numerous issues

[5] John Watkinson, The Art Of Digital Video, Focal Press 1994

[6] International Broadcasting, numerous issues

A number of other magazines, brochures, catalogues and books, and comments from readers (thanks to you people!)

Copyright © 1995-99 Mika Iisakkila. Reproduction of this document or any parts of it in any other form than linking to it via WWW is forbidden without permission from the author.

Back to home page