Frequently Asked Questions About DVD
by Kilroy Hughes
Future Media Systems
May 20, 1996
DISCLAIMER
This FAQ is presented "as is" and the views expressed in it are those of the author and are not necessarily
endorsed by The CD-Info Company, Inc. Please direct any
comments, corrections or suggestions to Kilroy Hughes, the FAQ's
author.
Author's Disclaimer: The information contained in this document is unofficial information from
public sources and possibly bogus personal opinions. It does not necessarily represent the official
specification or plans of the DVD Consortium, and none of the information was written subsequent to
disclosure under NDA of the official specifications.
Home | Applications | Bibliography | CD-Lite
History | Industry | Sponsors | Technology
Return to top
DVD stands for "Digital Versatile Disc". It includes products and software that will be built in conformance with
a specification being developed by a consortium of the largest computer, consumer electronics, and
entertainment companies. The intention is to create a range of compatible products based on a new
generation of the Compact Disc format which provides increased storage capacity and performance,
especially for video and multimedia applications.
It was just announced that the specification for video applications, referred to as "Book B" will be released
designated as version 0.9. There has been delay and reported disagreement between factions in the
Consortium over issues such as copy protection and regional release control for video movies that have
prevented agreement on a final 1.0 spec. Choosing to issue a preliminary specification could indicate that
some companies intend to launch computer products that are not dependent on movie industry agreement,
or just optimism that an agreement will soon be reached. The original plan of releasing products for
Christmas '96 has been seriously jeopardized by the delay, and companies are forced to decide right now if
they will have product at Christmas or wait for a consensus spec.
The Physical Layer and File System portions of the specifications seem to be well determined. These apply
to Book A, DVD-ROM; Book B, DVD-Video; and Book C, DVD-Audio. In the case of DVD-Video and DVD-Audio,
there is also an Application Layer specification which defines the software application and the functions built
into video and audio players. The video application layer is the current area of disagreement. The audio disc
application layer is in early discussion phase, and will probably receive low priority until video and computer
products are successfully launched. The audio portion of the DVD-Video spec is not in question. The
DVD-Audio disc format will be something new which uses the large DVD disc capacity for a higher audio
quality level than present audio CD's (for all those people who want 90 dB S/N at 30 kHz, I guess).
The specification is available from Toshiba for $5,000, and the execution of a very stringent non-disclosure
agreement. Licensing arrangements have not be announced yet. For CD-Audio, license fees were paid by
disc pressing plants, equipment manufacturers, and software developers who purchased Red Book
specifications. Mail me if you want the information to order specifications.
Return to top
There are five DVD "Books", labeled A through E pertaining to different applications. They are: Book A,
DVD-ROM; Book B, DVD-Video; Book C, DVD-Audio; Book D, DVD-WO (Write Once); and Book E, DVD-E
(Erasable or re-writeable). There is also discussion of DVD-RAM, which falls under Book E, but implies a very
high degree of re-writeability, like MO and phase change drives, and would be intended to function like a
removable hard disc.
The disc data format used for books A, B, and C is called UDF Bridge, which is a combination of "Micro UDF"
and ISO-9660. The "Universal Disk Format (tm)", or UDF(tm) specification was created by OSTA, the Optical
Storage Technology Association as an implementation of ISO/IEC 13346 standard. It defines data structures
such as volumes, files, blocks, sectors, CRC's, paths, records, allocation tables, partitions, character sets,
time stamps, etc.; and methods for reading, writing, and other operations. It is a very flexible, multi-platform,
multi-application, multi-language, multi-user oriented format that has been adapted to DVD, and made
backward compatible to existing ISO-9660 operating system software. Actual utilization of this disk system on
DVD discs will depend in large part on what Microsoft dictates as the operating system standard. Movie
players are supposed to use UDF, while computer applications will use ISO-9660 until UDF support
becomes universal.
The video standards being used include MPEG-1 (ISO/IEC 1117-2) and MPEG-2 (ISO/IEC 13818-2), with
quality levels ranging from fixed rate MPEG-1 at 30 fields per second at a resolution 352 x 240; to variable bit
rate MPEG-2 at 60 fields per second at a resolution of 720 x 480.
Audio standards include MPEG-1 (ISO/IEC 1117-3) stereo, MPEG-2 (ISO/IEC 13818-3) 5.1 and 7.1 surround,
and Dolby AC-3 5.1 surround and stereo ProLogic. Linear PCM at up to 24 bits per sample and 96 ksps in
stereo is also specified. Audio sampling rate for MPEG-2 and AC-3 is defined as 48 ksps, not 44.1 ksps as in
MPEG-1.
Multiplex streams conform to the MPEG-2 System Layer specification (ISO/IEC 13818-1). The average bit rate
assumed for a movie with three sound streams is 4.69 megabits per second. The DVD format transport
stream can deliver a variable bit rate of up to 11.08 megabits per second of user data containing up to 10.08
megabits per second of content and navigation data.
Return to top
The introductory product for DVD was supposed to be movie players introduced by several manufacturers
around Christmas '96. They are intended to cost from $500 to $750 and be supported by at least two-hundred
movie titles. Shortly after that, DVD-ROM drives for computers are supposed to be released, costing
anywhere from $200 to $500 for a basic drive; and an additional amount for a computer board to do MPEG-2
decompression, or built-in MPEG-2 electronics that will allow a DVD-ROM drive to be moved to a TV and used
as a movie player. At this point, the movie player launch for Christmas time is questionable, and might result
in an earlier DVD-ROM launch, possibly for Christmas. Unit sales estimate from manufacturers for DVD-ROM
drives range from 25 million/yr (Philips) to 50 million/yr (Pioneer) in the year 2000.
There have been various PR announcements saying that Sony Playstation, Matsushita 3DO (M2), Philips
CD-i, Apple Pippin, Sega Saturn, etc. all "plan" to come out with a DVD version of their existing CD systems.
Some have qualified this by saying "once DVD is successfully launched", which can be interpreted any way
they want, but is likely to take one or two years. Pioneer even has a Laserdisc/DVD player planned (at about
the price of two separate players).
"Entertainment PC's", such as the Destination PC, could be another important product type for DVD. These
are anticipated devices that would combine elements of today's multimedia PC's with platform video games,
"net computers" for internet access, interactive television set top boxes, and DVD video and music players. An
associated trend for all these video oriented products is a move toward the TV monitor, while at the same time
the "TV" is likely to go through major changes as it adapts to becoming a digital video display device resulting
in new product possibilities. TV limitations, such as poor text display, have been impossible to overcome
because of the design of the analog broadcast standards (NTSC, PAL, SECAM). Digital input monitors can
use methods common in computer displays, such as field buffering and progressive scanning to eliminate
flicker, interpolated pixel doubling to display at 1440 x 960 non-interlaced pixels; and new display
technologies such as flat panels and micro-mirror projectors to make both small and large displays more
effective.
Another product area is based on the small size and large capacity of the 3 inch (80 mm) DVD disc. It has a
1.4 gigabyte capacity (2.6 gigabytes dual layer), and will be re-recordable in the future. The small size and
cost could lead to its use in various portable devices and appliances, in preference to memory chips or hard
drives.
Return to top
The DVD specification is controlled by a group of ten companies referred to as "the Consortium". These
companies are some of the biggest consumer electronics, TV, computer, and entertainment companies in
the world, typically in the 50 billion dollar a year business volume range. Their accumulated annual budget is
comparable to the US government, so their ability to bring DVD to market is formidable. All the companies are
Japanese, except for two European companies and one American company. They contribute patents and
other technology to the design and implementation of DVD, agree to conform to the standard, and receive a
share of the licensing revenue that will be collected by an administrative entity to be designated by the
consortium.
The leaders of the group have been Philips and Sony, who developed and licensed the Red Book Audio CD
format; Matsushita, who joined Philips and Sony in the development of the White Book Video CD format; and
Toshiba and Time Warner, who focused on the needs of the movie and home video market. Pioneer is also
important as the main manufacturer of existing analog laserdisc technology. Hitachi, JVC, and Mitsubishi are
also included. Thompson/RCA/GE is also a member now, and has already made over 2 million MPEG-2
decoders for its DSS receivers. A year ago it looked like there would be two formats and a battle in the market
place that would leave nothing but losers. At this point there is a rocky agreement to agree on a unified
standard that would best serve the needs of all the companies and all the markets.
It is worth noting that this group represents most of the world's audio CD, TV and VCR manufacturing, with the
notable exception of some Korean companies; while the standard setters of the computer industry, Microsoft,
Intel, IBM, and Apple are all absent, but do have a technical interchange committee that has been a strong
influence. Representation of communication companies is also lacking, such as Telcos, Cablecos, DSS,
Netscape, etc., and these companies are probably viewed as competitive content delivery providers, rather
than partners in some vision of a network friendly DVD player/set top box using the TV for the full range of
anticipated multimedia applications. Nintendo and Sega, from the platform video game world, are also
absent.
Return to top
The storage capacity of a single sided, single layer, 120 mm (5 inch) DVD is 4.7 gigabytes. This is often
equated to 133 minutes of movie play time, assuming 3 audio streams and an arbitrary video quality/bit rate
of 3.5 megabits per second. It is made from two .6 mm thick substrates (discs) bonded together to form a
disc that is equal in size to a standard CD. If the second substrate is molded with information, the capacity is
doubled to 9.4 gigabytes (probably requiring a manual disc flip for full access). The manufacturing methods
for single layer discs, both single and double sided, are very similar to standard CD molding, with the addition
of special equipment for the Laser Burner Recorder and disc bonding.
A second layer can be recorded on either side using optical techniques and a photo-resin on top of the
molded layer. A 120 mm (5 inch) dual layer disc has a capacity of 8.5 gigabytes on a side, for a maximum of
17 gigabytes. An 80 mm (3 inch) dual layer disc has a capacity of 2.6 gigabytes on each side, for a maximum
of 5.3 gigabytes. The manufacturing method for the second layer is different from standard CD molding,
using UV cured resins and partially reflective/transmissive materials. The simplest method, for single sided
dual layer discs, is to bond two pressed substrates with UV resin, with one data surface coated with standard
reflective material and the other with partially transmissive material. The other method is to coat a finished
data surface with UV resin, press it, cure it, and coat it with partially transmissive material. The materials and
tolerances for dual layer discs are very stringent, and the playback equipment must discriminate between
reflections from the top and bottom layers that are not much different in intensity. It may be a while before
dual layer discs become common.
The nominal constant linear velocity of a DVD system is 3.5 to 4.0 meters per second (slightly faster for the
larger pits in dual layer), which is over 3 times the speed of standard CD, which is 1.2 mps. The nominal
transfer rate for application data of a DVD system is 10.08 megabits per second, about 7 times the base rate
of 1.4 megabits/s (172 kBps) for Mode 2 CD-ROM, and about 8 times the base rate of 1.2 megabits/s (150
kBps) for Mode 1 CD-ROM. The increased transfer rate is made possible, in part, by pits size of .4
micrometers and track spacing of .74 micrometers that are both about half that of standard CD-ROM. These
can be read because of the use of a visible red laser (650 or 635 nanometer wavelength), instead of the
longer wavelength infrared laser used for standard CD (780 nanometers). This change of laser wavelength
has the unfortunate side effect of making all the existing CD-R discs unreadable on DVD drives
(manufacturers are coming out with CD-R 2 recordable CD's for existing CD recorders that have dyes which
will work in DVD players and standard CD players).
Return to top
The data format for DVD Video discs allows for one video channel, but a large number of Program Chains
that can access it in various ways. There can be up to 8 audio streams that can be up to 5.1 channel's each
of AC-3 or MPEG-2 audio (or 7.1 if MPEG-2), or linear PCM and MPEG-1 stereo can also be used. Subpicture
data, which are four color graphics that can be used for subtitles, menus, etc., can be included in up to 32
channels which the user or title can select to play one at a time. There can be up to 99 title sets on a disc,
with up to 999 Part_of_Title segments. A typical combination of a 3.5 megabit per second MPEG-2 movie,
three sound streams at 384 kilobits per second, and a few subpicture channels adds up to about 4.7
megabits per second and results in a play time on a single sided disc of about 130 minutes; which is
supposed to be sufficient to hold 95% of the movies made.
Return to top
At this time, there are two separate applications for DVD; DVD-ROM and DVD Video. DVD Video describes the
disc, the data format and contents on the disc, and the play back system. DVD-ROM only describes the disc
and its format. The data and playback system are open to infinite variations, for good and bad. This is very
analogous to existing audio CD's (also Video CD and CD-i), which are system specifications; and CD-ROM's,
which are only a media specification. The advantages to a system specification are evident in the fact that
over 600 million CD audio players have been sold, and ten Billion discs: They are cheap and all those discs
work on all those players. The advantage of a media specification is that it can immediately replace other
media in computer applications, and it is flexible enough to adapt to the rapid evolution of computer
equipment and applications.
The basic capabilities that DVD-ROM brings are a large storage capacity, from 4.7 gigabytes to 17 gigabytes;
a fast transfer rate, about 8 times the nominal standard for CD-ROM (which is actually common for new
CD-ROM drives); and the potential for faster access speed and layer to layer access jumps (CD-ROM's will
also improve in this area with the introduction of constant angular velocity drives with 80 ms random access).
Additional optional capabilities include duplicating any or all of the DVD Video player's capabilities with add on
hardware or software. Some manufacturers have planned to sell MPEG-2 and Dolby AC-3 decoding boards
for PC's. Others have planned to build the circuitry into DVD-ROM drives, or make dual purpose portable
DVD-ROM/DVD Video players. These plans are based on the assumption that someone is making DVD
Video discs, and people will pay to play them on computers. At this point it isn't definite when those things
will happen.
However, it is certain that there will be another ten million computers capable of playing MPEG-1 video sold
this year, and most new machines will have the processing power to play MPEG-1 video in software from now
on. DVD gives these existing PC's the ability to play hours of MPEG-1 video at a high bit rates, which results in
very good video quality, "CD quality" audio, and backward compatibility to the thousands of Video CD's which
exist (a few hundred in the US). The difficult development process for DVD Video format using variable bit rate
MPEG-2 encoding, multi-stream multiplexing, cross platform porting, and emulation testing may convince
developers and publishers to wait a while before committing to MPEG-2 titles. DVD-ROM titles are likely to
contain the same old media types we are used to (AVI, QuickTime, WAV, etc.), with an increase in the amount
of MPEG-1 video, which is made easier by DVD storage capacity and speed, and the anticipated high
performance of newer PC's that would tend to be used with a DVD-ROM drive.
Return to top
One intended capability is wide ranging compatibility. DVD players should be able to play any video disc on
any TV system, PAL or NTSC, 5:4 screen or 16:9 screen. Players are also supposed to be compatible with
Audio CD's and Video CD's. Other CD formats, such as Photo CD, CD-i, Sony Playstation, 3DO, etc. will be
the option of the manufacturer. The intention is to appeal to users by combining the functions of presently
separate audio and video entertainment machines (VCR and Audio CD) into a single entertainment player
that can play the existing discs. It also takes advantage of the trend of integrated TV/multichannel audio
"home theater" systems, which combine the audio and video presentation systems, and are ready for
multichannel sound.
DVD Video is designed with international compatibility in mind. It is capable of eight audio tracks (although
this would use over 3 megabits/s of the disc's bandwidth, leaving no space for normal video), and 32
"subpicture" tracks (used for subtitles, menus, etc.); which can be used to put several selectable languages
on each disc.
The DVD Video spec (Book B) describes the disc, the data format and contents on the disc, and the playback
system. The advantages to a system specification are evident in the fact that over 600 million CD audio
players have been sold, and Billions of discs. A standardized system is cheaper to manufacture because all
models are similar and designed to do a limited number of things as simply as possible; and they all work
using any disc with any player. That is in contrast, for instance, to the current situation with CD-ROM's on
PC's, where the majority of new titles don't work on the majority of existing computers. The DVD Video
specification is in flux right now with at least some of the universal compatibility, that was a feature of CD
Audio being sacrificed for the business interests of the movie studios. At the least, there is a different audio
encoding standard for NTSC countries (Dolby AC-3 5.1 surround), and PAL/SECAM countries (MPEG-2 audio
5.1 and 7.1 surround). There will probably also be playback limitations based on geographic regions and
possibly date that will prevent discs from playing at a place and time that doesn't conform to the movie
studio's release schedule.
Return to top
For maximum compatibility with different content material and display monitors, every player has the ability to
select aspect ratios to best fit the monitor and video material being watched. Most movies have an aspect
ratio (width/height) of 1.85, and some anamorphic (which means, like the human visual field) wide screen
movies have aspect ratios of 2.25. Standard TV programs and TV sets have a ratio of 1.33 (4/3) (...and
640/480 computers do also). There are new TV's with 16:9 screens (aspect ratio 1.78) that can display a
standard size movie by chopping 2% of the width off each side. In other countries than the US, these wide
screen TV's are becoming common. More than 30% of the TV's sold in Japan in the last year have been wide
screen. For standard width TV's or any TV showing an anamorphic movie, the options are to "pan and scan",
squeeze, or "letter box" the movie. Squeezing looks silly, with everyone looking twelve feet tall. Pan and scan
will work the best sometimes because it can fill the screen with objects of interest, such as peoples' faces;
but it has many problems, like when people are standing a distance apart in the movie and one of them
doesn't fit into your picture. DVD has special provisions to contain "center of interest" coordinates to tell the
player how to crop the picture dynamically if you select to view the video in pan and scan mode. If you select
letterbox mode, the scan lines are combined together so that every 4 lines results in 3 and a 360 line movie is
displayed on the 480 line visible screen. There is a loss of resolution with a smaller image and an unused
band of black above and below the picture, but the original composition of a 1.85 movie remains intact, and it
is the best way to view most movies shot outdoors, or with crowds, or with good photography.
Advanced TV displays will be possible using the digital output from DVD Video discs. For example a
DVD-ROM drive with decoder can supply digital video data to a display system that buffers both fields and
uses a high refresh rate and progressive scan for the display, or directly addressed pixels in a flat panel or
micromirror display. Increased resolution by line or pixel doubling is also possible using calculated
interpolation and scaling such as many computer displays use to display full screen images from small
MPEG or AVI videos. DVD provides a path for continuous improvement of video display quality that has been
lacking with fixed analog systems or the monumental task of switching the broadcast system to a new HDTV
standard. Display improvements will enable the TV to handle many applications that have been restricted to
computers such as those that require lots of text display.
Return to top
The interactive capabilities of DVD players are limited to menus with a small set of navigation and control
commands, with some interesting functions for dynamic video stream control, such as seamless branching
(which can be used for playing different "cuts" of the same video material for dramatic purposes, censorship,
etc.). Commands can be imbedded in the video stream and menus, but there is no real programming
capability like a computer, or memory access (other than a few registers), or rendering capability for cursors,
text or graphics. The user interface is limited to number and arrow buttons on a remote control, with the main
GUI being four color "subpictures" overlaid on video or video stills with the ability to step sequentially through
menu selections which highlight when selected. Because it lacks a cursor to select an option by pointing, or
find subtle "hot spots" which respond with text, animation, audio, etc. when touched; and because four colors
are inadequate for pictures, the DVD interface is mainly designed for numbered text on a menu that you select
with number buttons ... just like TV channels!
It is possible to construct "computer-like" applications with logic branching and pause for user input. This
could be used for simple training and other applications. For example, a disc could play a video segment and
ask a question, pausing on the last frame. A menu could be displayed consisting of a few numbered
answers created with a subpicture overlay. When the user picked a number or the menu timed out, the
selected menu command would play the associated PGC (ProGram Chain) which defines a sequence of
video to play, which could be remedial, or "on with the course" if the answer was correct.
The following is a quick description of most of the other interactive and control functions of the DVD player.
There are 16 General Parameters available to the title to store information, like scores, book marks, user
information, calculations, or menu selections. Three types of still play are provided for, including paused
video frames at the end of a PGC, or the end of any Cell in a PGC (like a slide show), and a Video Object Unit
still which is a still picture not encoded as motion video. Navigation commands include simple arithmetic,
Button Commands (up to 36), GoTo, Link, Jump, Compare, Set General Parameter, Set System Parameter.
There are 24 System Parameters that store specific data like audio, video, and subpicture stream numbers,
timers, language and country codes, camera angle, and "Parental Level" for censorship. There are several
Scan functions; by time, video sequence (backward or forward), and by branch. Discs are organized into Title
Sets, Part_of_Title (which can be scenes, songs, chapters, acts, innings, etc.), Program Chains, and Video
Objects.
Return to top
Program Chain Information Files (PGCI) are lists of pointers to video streams, and also contain embedded
commands. Normally PGCI's are played sequentially, but they can be played in random or shuffle sequence
(like songs on a CD player). The video segments addressed by a PGCI can be accessed at the GOP (Group
of Pictures) level, can be non-contiguous and non-sequential, and may be contained in part or in whole in
other PGCI's. This is possible to a limited degree based on the player's "seamless jump" capability.
Seamless video jumps can be made within the limits determined by the size of the track buffer, which holds
the User Data coming from the disc at 11 megabits per second, the Variable Bit Rate of application data,
which draws data from the buffer at up to 10 megabits per second, and the maximum seek and spin latency
time that will not cause the track buffer to run empty and cause a video error, which is a function of the disc
layout and drive mechanics. Video Title Set Information includes control data such as a Time Map Table,
Part_of _Title pointers, and Menu and Title Attributes. Display attributes include fade, scroll, transparency and
highlighting, and are used for presenting menus, captions, and graphics.
Rapid frame rate play of subpictures allows a crude form of prerecorded animation playback. The four on
screen colors of a subpicture are selected from a palette of 16 contained in the PGCI. They are designated
Foreground, Background, Emphasis-1, and Emphasis-2; or the emphasis colors can be used to provide a
crude level of antialiasing. Subpictures can be up to 720 x 480 (NTSC) resolution, but they must have a run
length encoded size per line under 1,440 bits and fit in a 62 kB buffer. The data structure of a subpicture
includes a Display Control Sequence Table (SP_DCSQT) that controls the presentation of the pixel data with
any number of Display Control Commands (DCC); and subpicture content, color, contrast, and position can
be changed as often as every video field. The subpicture system is perfectly designed to scroll movie credits
and do captions in 32 languages that defy embedded rendered fonts (consider a font generator for 10,000
Chinese characters ... in 17 different styles.); as opposed to producing multiple video versions in each
language for titles, credits and captions.
Return to top
MPEG-2 video conforms to the ISO/IEC 13818 specification for encoded video data and the multiplexed stream
containing audio and other information. The emphasis for the DVD movie player is on high quality, 720 x 480
(NTSC), 60 field per second video encoded from 24 frame per second film using variable bit rate encoding
that avoids compression artifacts on difficult sequences by using higher bit rates, and reduces bit rates for the
easy sequences, typically averaging around a 30% reduction in bandwidth and storage capacity compared to
fixed bit rate. The video standard for DVD is actually very flexible and allows for fixed bit rate encoding,
MPEG-1 or MPEG-2, and lower resolutions like Video CD resolution of 352 x 240. MPEG-2 video is used in
other systems such as satellite broadcast video and interactive video over phone lines, but these
implementations do not conform to the DVD video spec, which is a subset.
Both MPEG-1 and MPEG-2 rely on discrete cosine transform intraframe encoding followed by interframe and
motion prediction encoding. The amount of computation required for standard DVD MPEG-2 video decoders
is the equivalent of multiple Pentium computers, encoders require several times more. MPEG-2 encoding
systems presently cost hundreds of thousands of dollars. Specialized encode and decode processor chips
are required to supply the calculation speed. At present MPEG-1 decoding in software is possible on newer
Intel CPU's used in PC's. In the future, faster CPU's with MMX very long instruction word integer processing
may be capable of doing MPEG-2 video decoding in software, but for now specialized decoder chips are
required. This means that today's computers with a DVD-ROM drive added won't be able to play DVD movie
discs, unless an MPEG-2 and Dolby AC-3 video and audio decoder board is added to the PC or built into the
DVD-ROM drive. However, MPEG-1, AVI, QuickTime , and other media formats will play from the DVD-ROM as
though it was a very big 8X CD-ROM drive.
The playback quality of DVD video can be as good as a D1 master tape, but it depends on how much
bandwidth is used to encode the video, and the nature of the source material. Since motion prediction and
interframe encoding use 16 pixel square macroblocks for calculations, inaccuracies in encoding result in
visible macroblock outlines or rapidly changing patterns, sometimes called "mosquitoes" in areas that should
be a solid unchanging color. Image sequences of many small random changes require the most bits to
represent, and certain regular patterns will make errors more visible. Film is a good source because it is high
resolution, low noise, and usually has only 24 non-interlaced frames per second, which allow more bits to be
used to encode each frame, compared to 30 frame per second sources. Video sources recorded in
component formats, like D1, Digital Betacam, DVC and Beta SP give the best results, with digital formats
generally being better. Some digital compression or processing used in recording or editing creates
changes in the video that may not be visible, but cause major problems in MPEG encoding due to many small
changes that the encoder tries to encode. Similarly, noise or analog distortions such as ringing or saturation
on a video tape will consume large amounts of bandwidth and lower the relative quality.
Return to top
Encoding variable bit rate MPEG-2 video is much more complicated and expensive than more common
MPEG-1 encoding. MPEG-1 audio/video can be encoded and multiplexed in one pass in real time, which
allows compressionists to make fine adjustments to optimize the encoding while they observe the results.
MPEG-2 requires multiple passes to encode the video, additional slow passes on each audio stream,
storage of many gigabytes of elementary streams, editing all streams for synchronization, mapping program
chains and jump points, an authoring process to create several types of video objects, subpicture streams,
display and navigation commands, etc., and a multiplexing process to combine all this data into a single
complex stream. An emulation system has to be used to play the resulting multiplex stream, or a disc can be
mastered and replicated at a pressing plant. This process may take a week or more, and if there is an
adjustment required, some or all of the process will have to be repeated to see the result.
DVD recorders might speed the disc development process by acting like a quick pressing plant for check
discs, but they would only be capable of recording from an existing premaster image made from the multiplex
stream described above. Book D on DVD-Write Once is considering a 3.9 gigabyte recordable disc. This
would be significantly different than a 4.7 or 8.5 gigabyte disc and would be of limited value in performance
testing before production. The prospect of a real time "DVD movie player format" recorder is not on the
techno-horizon because of the multi-pass nature of variable bit rate recording, and the authoring and
multiplexing required. A simplified, fixed bit rate, linear video DVD recorder could be made, but even with
MPEG-1 encoding it would probably cost over ten thousand dollars for the next couple years: Hardly likely to
take the place of VCR's or $500 CD-ROM recorders. The most common use for DVD-WO recorders will
probably be as computer data recorders for DVD-ROM discs, because any encoding, authoring, and data
formatting can use existing PC formats, and precise disc layout is rarely a concern on PC's because their real
time performance is inconsistent from one system to the next that close tolerance real time disc performance
is rarely relied on anyway.
Return to top
The standard calls for all DVD Video discs intended for countries using NTSC format to use Dolby AC-3 2.0 or
5.1 channel surround sound at 48 ks/s (5.1 refers to a center channel, left front and rear, right front and rear,
and a low frequency effects channel, or sub woofer, which rates "point one"). Discs intended for PAL/SECAM
format must use MPEG-2 or MPEG-1 audio format at 48 ks/s with 2.0, 5.1 or 7.1 channels (MPEG-1 stereo
data is contained in MPEG-2 streams). The other non-required audio format in each case can be used in
addition, but this can be a practical problem because of storage space and bandwidth limits. Either type of
disc can also add a linear PCM stereo audio track with a sample rate of 48 ks/s or 96 ks/s, with 16, 20, or 24
bit samples, not to exceed 6.144 megabits per second. Each five-channel stream uses a bandwidth of 384
kbit/s; if all eight streams were used for multichannel recording of sound tracks in different languages or both
audio standards, 3 mbit/s would be consumed, and not much left for video. It is possible for a maximum of
seven voice tracks in different languages to be recorded separate from a music and sound track multichannel
stream. It may be possible for voice streams to be recorded at 64 kb/s in mono, or at 128 kb/s in stereo and
mixed during playback based on the user's selection. The resulting bandwidth consumption would be
reduced to as little as 832 kb/s (for center channel mono).
It remains to be seen if players will be equipped to play back both formats or discs will be made with both
formats. It is possible that the two different audio standards will effectively create two incompatible types of
DVD disc. (Compatible NTSC/PAL Video CD's have been used for years. The biggest problem is the loss of
about 50 lines top and bottom when a video designed for PAL only playback is played on an NTSC TV.) This
audio standard incompatibility may be the result of the established position of Dolby AC-3 and Dolby ProLogic
in movies in the US, or the desire of movie companies to keep their product from showing up in PAL countries
until they release it there. Any computer that has MPEG-1 playback will be able to decode stereo from any
MPEG stream. A computer will require an add on processor to decode AC-3, and special provisions to mix
5.1 channels into a stereo signal.
Digital audio outputs will be standard on DVD players. Both coax and optical outputs in IEC-958 format,
similar to DAT and some CD players, will be normal. Amplified speakers will be particularly attractive for new
systems, because a single cable can be run past all the speakers, and they can tap the correct signal for their
position. Most existing audio/video amps that are built for stereo and ProLogic inputs won't be able to handle
5.1 or 7.1 amplification. Another capability that some amplified speaker systems may have is to adjust the
time delay for each speaker to adjust for room and placement variations. This is a small part of the difficult
task of recording, mixing, and reproducing five or seven channel sound so that it reproduces the original
acoustic environment of the source, or creates some kind of coherent three dimensional experience.
Return to top
Aside from increased DVD storage capacity that results from tracks that are half as thick and pits that are half
as long, DVD's have a multilayer capability defined that allows a second partially transmissive layer of slightly
smaller capacity to be printed on top of the CD like reflective data surface. The laser reader is designed to
adjust its focus to either layer depth so that both of them can be quickly and automatically accessed. It is also
possible to print the other side of the disc to double its capacity because it is made from two platters of half
thickness that are bonded together, resulting in the same dimensions as a CD disc. A shorter wavelength
laser is used to read the smaller geometry, but special lenses are used to give the DVD reader the ability to
read pressed CD's also. At present, the DVD readers won't be able to read CD-R discs because the different
laser wavelength is not reflected properly by the CD-R dyes in use now. New CD-R blanks with new dyes will
solve that problem in the near future.
Beyond the physical properties of DVD discs, there are differences in format and data. DVD's use a disc
format called Micro UDF Bridge Format. This format combines a new, very flexible adaptation of the UDF or
Universal Disc Format (tm) with PC CD-ROM's disc format, ISO-9660. While the data format in files on
DVD-ROM's is undefined, like it is on CD-ROM's, Book B of the DVD specification defines a DVD Video Disc
application, and consequently the characteristics of a DVD Video Player. These DVD Video discs are very
different than CD-ROM's, but similar to Video CD, CD-i, Audio CD, and CD-ROM XA in that they all contain
multiplexed data streams that must be delivered in real time, three of these formats multiplex audio and
video, and two of them use multiplexed MPEG streams, just like DVD. While DVD-ROM and DVD-Video discs
are physically similar, the process of authoring the DVD-Video disc multiplex stream is very different than
copying a PC application's files to a premaster tape to make a DVD-ROM.
The good news is that great effort has gone into making the DVD-ROM drives backward compatible. That
means you can buy one knowing that you will be able to at least play the existing CD format if new DVD-ROM's
don't show up in big numbers any time soon. The alternative in buying a new drive is to get one limited to the
old CD format and possibly find that it is obsolete soon because the software you want is only on DVD. The
biggest glitch for now is that any of the existing CD-R discs aren't backward compatible on a DVD-ROM drive.
Another question is what level of performance DVD-ROM drives will have playing a CD disk. The playback
performance of a DVD Video disc is defined, but the speed it plays a CD-ROM (6X? 8X?) is not. The spin rate
required for DVD purposes is only about 3X, because of the higher pit density, so high speed CD-ROM
playback may be an extra cost.
Return to top
The focus on software has been movie discs intended for DVD Video players. Time Warner has publicly
stated that they intend to have 250 movies ready for the launch of the players. Other movie/TV studios such as
Sony, MCA, MGM, and Turner have said they will release DVD's also. Disney is one important home video
studio that said they won't, because of concern over copying. Paramount may hold out because of their
connection to Blockbuster. The issues that might hold up the movie software are lack of copy protection and
regional release control that the studios think is satisfactory. Whatever is done with movie players to add
those protections seems to be bypassed by DVD-ROM drives that can do anything under software control,
even read the raw data from the disc to make copies. The computer industry isn't willing to subject computer
peripherals to the control of the movie industry, so agreement has been hard to come by, the specification has
been delayed, and movie software will be delayed until some agreement is reached.
The kind of movies that are targeted for DVD may be different than the tape rental business. Tape rental and
cable are mainly used for one time presentation of new releases at some time delay after their theatrical
release. The hope for DVD is that discs will be purchased like books and music CD's and watched several
times. The studios look forward to issuing "classics" from their back catalogs, assuming that people will
have some number of movies they've seen in their life that they think are important enough to own, and that
new content and interactive features will make it more attractive for people to watch movies more than once.
The idea that people will buy the discs is also supported by the fact that video tape sales will be about $15
Billion this year, with eight tapes per average buyer.
The production process for DVD movie discs is relatively long and difficult at this stage of development. A
16:9 film to video transfer and a virtual pan and scan version have to be produced for this high resolution
medium (in PAL and NTSC). This is an esthetic process, not just a mechanical one, which requires directors
and producers and time. Multiple multitrack audio mixes may be needed. Dialog may need to be translated
into several languages and caption graphics made. Interactive features have to be designed and tested. New
content, special edits, parental control versions, interviews, background information ... many elements in
addition to the mechanics of MPEG-2 encoding, disc authoring and mastering will slow the release of movie
discs. The tendency will probably be to release simple linear movies for a while before attempting to routinely
use some of DVD's more interesting capabilities.
Return to top
In the computer world, DVD-ROM software should arrive by the shovelfull. Any PC files can be stamped on a
DVD-ROM by copying them to a tape and sending them to a pressing plant. Existing software will play on
existing PC's from a DVD-ROM drive just like a 8X CD-ROM drive, with the main difference being a larger
storage capacity. This makes MPEG-1 software work particularly well because there is plenty of storage for
the large files, there is plenty of bandwidth to use high bit rates and get excellent video quality, and encoding
and decoding on PC equipment is becoming widespread and inexpensive.
DVD-ROM's using MPEG-2, including movie discs, present problems for PC playback. Most DVD-ROM drives
bought will probably not have MPEG-2 video and AC-3 audio decoding built in, and none of today's PC's will be
able to decode it in software. Over the last few years, only a half million MPEG-1 boards have been sold for
PC's. It is unlikely that large numbers of computer owners will buy expensive MPEG-2 decoder boards so
they can watch movies in six channel surround sound on small computer monitors. It is unlikely that
computer software developers will go through the great expense, limitation, and difficulty of making MPEG-2
DVD's that will play on movie players, unless a very large installed base develops quickly. If they rely on
MPEG-2 video on a DVD-ROM (only) disc, they will have a very small number of computer owners who can
play it back. If they use MPEG-1, there are about 10 Million computer owners who can play it (IF the computer
owners buy DVD-ROM drives). On the other hand, software released on CD-ROM will play on the small
number of DVD-ROM drives sold in the next couple of years (some estimate 2 million), and also on the 65
million CD-ROM drives estimated to be in use world wide at the start of 1996, about 38.7 million of which were
bought during 1995.
Return to top
Unfortunately existing CD-R discs won't play on DVD players because of the different laser wavelength used
for DVD. In the near future, CD-R 2 blanks will become available which are supposed to work in existing
recorders, CD-ROM drives, and DVD-ROM drives.
The prospect of a DVD video recorder is very unlikely in the existing DVD Video Disc format. A DVD-WO (Write
Once) is planned under Book D of the DVD specification, and prototypes have been demonstrated. These
would be data recorders with a capacity of 3.9 gigabytes per disc. They would function like today's CD-ROM
recorders. Several manufacturers have said that they think their DVD-WO is a couple years away from
product introduction. Pioneer says they plan to sell a DVD recorder in the middle of 1997. A potential
competitor for digital video recording that is much cheaper than recordable DVD is a digital video cassette
recorder. They are sold now in cameras, complete with "firewire" digital I/O, and could replace the VCR,
record digital video sources like satellite, DVD, cable, etc.; use cheap tape, and do most of what DVD is likely
to be used for on playback (Don't forget how laserdisc vs. VCR came out).
Another DVD product in planning, covered in Book E, is DVD-RAM, which is a disc and drive with a tentative
capacity of 2.6 gigabytes that can be rewritten many times. If the cost of these DVD recordable products is
high, they may have a slow rate of acceptance because there are cost effective write once and rewritable
products available now. (For instance, an Apex disc holds 4.7 gigabytes of rewritable storage with hard disk
access and transfer performance, for about $1,500. What will it cost in two + years when DVD-RAM is ready to
compete with it?) Philips and other manufacturers are planning to release rewritable CD-ROMS in early '97. A
compelling advantage for DVD-WO would be a large installed base of DVD-ROM drives that could read the
discs, and a low cost for blank media. That only becomes important when storage requirements regularly
exceed the capacity of CD-ROM discs.
Return to top
Developers in the CD-ROM area are faced with the option of publishing on CD-ROM or DVD-ROM formats;
and if they chose DVD-ROM, will they try to go cross platform with a disc that will play on DVD Video players?
One likely publishing opportunity will be bundleware that is sold with DVD-ROM drives and decoder boards. It
is a familiar mode that has been developed for CD-ROM, 3D and MPEG-1 (etc.) hardware. The hardware
manufacturer has to supply some software that works on his widget, or the buyer will be greatly disappointed.
This may lead to some challenging software intended to demonstrate the capabilities of the medium, paid for
by the manufacturer, but never likely to be profitable on the basis of units sold during the startup phase of
DVD.
Any publisher who hopes to sell DVD-ROM's to the largest possible market will have to include MPEG-2 and
AC-3 as just one option, if at all, because most DVD-ROM drive owners won't be able to decode it. Including
any video in MPEG-1, either in addition to MPEG-2 or instead of it, would open up the potential market for a title
to most of the DVD-ROM owners (assuming they have newer computers with MPEG-1 decoding, which is
likely).
Any of the CD-ROM's in distribution now could be put on DVD-ROM, but there would be no need to do it
because the CD-ROM's would already play on the DVD-ROM drives. Any publisher who released a title on
DVD-ROM would also need to sell it on CD-ROM to reach the majority of the market, so why not skip the
DVD-ROM version. Since the CD-ROM would play on the DVD-ROM player, the main reason to publish the
DVD-ROM in addition would be if the title exceeded the capacity of a CD-ROM disc, and it was cheaper or
easier to use one DVD disc than multiple CD discs.
Another area that gets surprisingly little attention is music. Just consider that there have been 10 Billion
music discs sold (multiples of the CD-ROM industry), 600 Million players sold, and every music performer has
lots of promotional music video's they've spent thousands on, but won't get a nickel for. What do people buy
and want to own, not rent? What do they play over and over? What is the largest application in the world for
MPEG-1 players? (Hint: It starts with a "K" and ends with a hangover.)
Return to top
Authoring for the DVD movie player is a very different process compared to most CD-ROM authoring. It
involves porting to an isolated playback device, a DVD player, which does not have computer-like storage and
access to allow native development and testing. It also relies on data combined in a multiplex stream of
synchronized digital video and audio, and other data and commands that behaves as a real time system
combined with the disc and player. Playback of content in development can be simulated with media files
prior to multiplexing using MPEG and AC-3 decoders, mastered and pressed onto a DVD disc and played in
a player, or emulated by a computer and hard disk imitating a DVD drive and disc and connecting to the
electronics section of a real DVD player. The process has some similarities to authoring for video game
platforms, CD-i, and Video CD; but it is very different from most CD-ROM authoring, which just copies a
working application to a CD-R without porting, complex real time multiplexing, or emulation.
From the start, a DVD Video production should plan for the interactive and display capabilities of the medium.
A 16:9 aspect ratio transfer to video may need to be created. A virtual pan & scan version of that needs to be
designed using "center of interest" coordinates to control the 4:3 viewing window. Special cuts need to be
edited for different rating levels (PG, R, etc.), and the sound track "dealt with" in some way that allows for
occasionally dropping chunks of the video in edited versions, or making interactive jumps to other video
segments. Multichannel audio needs to be scored, Foleyed and mixed to fit the various program chains and
their timing and spatial layout. Note: Truly innovative interactive video would have to start with the design and
production stage in order to capture various points of view on camera, create alternative plot sequences, etc.
for nonlinear story telling, games, performances, or things that haven't been thought of yet.
The MPEG-2 video encoding process involves working to a bit budget at the start that will tell you how much
bandwidth can be used by clips as they are encoded. Variable bit rate allows the compressionist to make
judgments about which segments will require more or less bandwidth so that the highest quality is achieved
within the storage and bandwidth limits. Allowance must be made for interactive jumps also because they
will effectively require a couple megabits per second of additional bandwidth (it is actually a function of the
disc layout, drive mechanics, and a player's track buffer, but the net effect is a reduction in the allowable
bandwidth). Encoding is a two pass process where the video is analyzed first to establish the encoding
parameters, and then the video is captured, compressed, and recorded to a big hard disk, typically 9 gigs,
through a RAM buffer. The audio process is more automatic, but it make take several times real time to
software encode on high speed workstations. Hardware encoders for MPEG-2 and AC-3 will improve this
situation in the future. A two hour movie could easily take several days to encode, preview, adjust, and
re-encode.
Return to top
Menu elements consisting of still pictures, video stills, subpictures, and supporting data structures and
commands need to be created. Data needs to be created and formatted for Video Manager Information, Video
Title Set Information, Presentation Control Information, Data Search Information, etc. Commands can be
embedded in menus, subpictures, Program Chain Information files, and video objects that include; Goto,
Link, Jump, Compare, Set System Parameter, Set General Parameter, etc.; and presentation control
commands to control object contrast, position, highlight, color, etc. These elements are created and
combined within an authoring system to implement a navigation map and user interface that will hopefully
operate the built in programming of the DVD Video player when a disc is eventually manufactured.
The audio, video, and subpicture data has to be edited to correct lengths and synchronized like a multitrack
recording, in preparation for multiplexing into a single data stream. The various streams are mapped into
channels that match the navigation commands, menus, etc. so that the interactive program can properly
access them. Program Chain Information Files (PGCI's) need to be constructed similar to edit decision lists
(EDL's) to describe the various "virtual video clips" that are playable from the encoded video content. All Jump
Points combinations need to be taken into account so that only legal jumps are possible based on the disc
layout and maximum allowable jump time. Title Sets, Part_of_Titles, Cells, Video Object Units, and Time Map
Tables all need to be designated or created.
As the media files are prepared and a layout created, it is necessary to use simulation testing to verify that the
presentation is acceptable. Simulation testing involves using a method of presentation that is different than a
DVD player to approximate the playback result that will be seen with the DVD player. It may be possible to
simulate functions such as navigation, user interaction, and timing based on the authoring information
accumulated and simple MPEG or AC-3 decoding of the elementary media streams. At a minimum it is
important to be able to simulate synchronized audio/video playback so the content of the disc can be validated
or re-edited and re-encoded.
The multiplexing process combines the various media streams, video objects, data objects, and control
commands into a single data stream that will be eventually wrapped in a disc file structure and burned onto
the disc master. This multiplex or transport stream is like the score of a symphony orchestra, the actual
performance where a bunch of incoherent bangs and toots come together to make Mozart.
Return to top
Emulation testing uses a computer set up to imitate a DVD disc drive and disc using its hard disk. A disc
image is created on a hard disk by a long complex computer process to resemble the layout of an actual disc.
An in-circuit emulator wires into the electronics of a DVD reference player to approximate as closely as
possible, the performance of the actual disc that would be made from that transport stream still in the
computer. An emulation program is started, and the title played through the DVD player. This may be the first
point at which an accurate evaluation can be made of the interactive aspects of a title. If something isn't right,
it might be necessary to go back to the post production or (heaven forbid) production stage and do it all over
again.
The final authoring step is to create a pre-master tape using the multiplex stream that has the correct format
for the pressing plant. The pressing plant will apply certain disc formatting as the contents of the tape are fed
into the Laser Burner Recorder (LB.), which makes the master disc.
Replication involves multiple stages that create one or more stampers that are used to injection mold
substrates, or print UV resin layers on multilayer discs. After the two substrates are possibly printed and
definitely bonded, you have a DVD disc that can be tested..
Since an emulator may cost hundreds of thousands of dollars and require engineers to operate, it is not likely
that it will be used for much user testing. Since extensive user testing will rely on DVD players and pressed
discs, any bugs or needed improvements found at this stage will required repeating some or all of the
process, which might take weeks to produce another disc. Did I mention that this is harder than CD-ROM?
Return to top
Home | Applications | Bibliography | CD-Lite
History | Industry | Sponsors | Technology
Copyright 1996 Kilroy Hughes
Email comments to kilroy@aa.net
HTML copyright 1996 The CD-Info Company, Inc.
960628 2037