Understanding MPEG-2 Compression

High quality video compression is critical for DVD because the data storage requirements for uncompressed CCIR-601 resolution 4:2:2 serial digital video are approximately 20MB/s (MegaBytes per second).  This means that a 2 hour long film in uncompressed CCIR-601 format would require approximately 144 GigaBytes of storage, and that is before accounting for any audio!

With DVD currently being capable of storing a maximum 4.7 Gigabytes of data, compression ratios of approximately 40:1 are required to fit the video for a feature film along with the audio and sub-titles onto a single sided 4.7GB DVD disc.To address this compression problem, the Motion Pictures Experts Group developed MPEG-2 video compression.MPEG-2 compression is based on the principle that temporal and spatial redundancy in motion pictures make up the majority of the visual information that humans perceive. (In simple terms there's a lot of information in an uncompressed frame or movie that we do not need)By comparing changes from frame to frame and removing as much of the "same" (redundant) information as possible the data storage and transfer requirements for the media are dramatically reduced.

An MPEG-2 stream is comprised of I-Frames, B Frames and P Frames

A standard sequence of [[I-B-B-P-B-B-P-B-B-P-B-B-P-B-B-I]] would have a Group Of Picture of 15, with 15 representing the interval at which I-Frames repeat.

Illustration of an MPEG 2 stream with a GOP of 11 How MPEG-2 Adjusts To The Complexity Of The Picture

I -Frames are the building block of the MPEG-2 sequence. B and P frames are based on the assumption that the video sequence will not change drastically between I frames, which would impair the B and P frames from accurately representing the motion of pixel information from one I frame to the next. In video production, however, rapid scene changes and excessive motion can have serious negative impacts on the quality of the video. Thus, the sequence of I to B frames must often be modified. This is accomplished by adding I-Frames. I-Frame "forcing", as this process is known, is usually done on scene cuts. In this manner, each new scene can start with a fresh I-frame which is then used for the calculation of subsequent B and P frames

The use of Variable Bit Rate In MPEG-2 Encoding

MPEG-2 is a scalable compression scheme in that the bit rate may be increased from a low 1.5 megabits per second to as high as 40 megabits per second. In the DVD format, the maximum allowable bit rate is approximately 10.5 megabits per second. The higher the bit rate, the more bits allocated to each frame and the better looking the picture. Additionally, to reproduce complex detail, motion etc, the Group Of Pictures sequence may be varied from a Group Of Pictures of one I-frame to a Group Of Pictures of 15. The lower the Group Of Pictures, the closer the location of I-frames and the higher the relative bit rate.By definition, I-frames contain much more information than P and B frames. As I-frames are inserted into the picture sequence, the data requirements and the bit rate for MPEG video increases. In order to place a 133 minute movie onto a single sided 4.7 Gigabyte disc, the average MPEG2 video bit rate must be kept as close to 3.5 megabits per second as possible By allowing for bursts of up to 10 megabits per second, the DVD specification takes into account the need for short bursts in the data rate to accommodate complex scenes where additional I-frames or higher overall bit allocation are necessary to prevent visible e compression artifacts. During less complex scenes, the video bit rate may actually fall to below 3.5 megabits per second, to make up for the early consumptive scene. Thus, variable bit rate (VBR) MPEG-2 encoding is critical for optimal picture quality. Ultimately, however, there is a trade off between perceived quality and total minutes of video and audio that can be placed on DVD. The use of a variable bit rate goes a long way toward improving the subjective quality of low bit rate MPEG-2 video, but additional work (noise reduction, flicker reduction, removal of grain in the picture) prior to encoding can also help.

DVD as a 'Bit Bucket'

A DVD disc is a data bank. It has a finite amount of information that it can warehouse. Depending on the disc construction, storage capacity varies from 4.7 gigabytes (DVD-5) to 17 gigabytes (DVD-18). To understand that massive storage, consider that the lowest common denominator is a single bit. Eight bits equal one byte. The DVD-5 has 4.7 billion bytes of information stored on it. That's the equivalent of 37.6 billion bits--on the single side of a DVD disc. That's lots of raw data, and it all comes down to how you manage that data, how you manipulate the retrieval of data from the disc to produce the highest quality audio and video images.Based upon an average bit rate of 4.7 megabits per second, a DVD-5 can yield up to 133 minutes of high quality (better than Laserdisc) video with 3 audio streams and 3 subtitles. A Dolby AC-3 audio stream requires 384 kilobytes per second; subtitle data requires 10 kilobytes of data rendered per second. At an average bit rate of 4.7 megabits per second, that leaves approximately 3.5 megabits per second that is dedicated entirely to reproducing the video image.

Complexity Revisited

The video bit rate of DVD varies continuously with the complexity of the image being created. Although the average bit rate for DVD is 4.5 megabits per second, the actual data rate can actually slip below 3 megabits per second and accelerate to 10.5 megabits per second. This is the method by which consistent, high quality images are generated from a fixed data bank (ie: the DVD disc). During high complexity scenes (fast motion, detail etc.) the bit rate is accelerated. To balance, or compensate for this massive withdraw of data from the bank, the average bit rate is reduced during low complexity images or scenes. In any given scene, therefore, the bit rate may vary to either extreme.

Quality and Storage

At an average bit rate of 10 megabits per second, a DVD disc can produce a picture quality that is nearly the equivalent of a studio master tape---the medium from which all pre-recorded software is produced--. If such a disc were to be produced, it would yield approximately one hour of audio and video per side, based upon a 4.7 gigabyte DVD disc.Similarly, if a producer wanted to put more than 3 languages or 4 subtitles on a DVD disc, or if the producer desired to incorporate higher bit rate streams (like DTS), they would require a disc with higher carrying capacity (perhaps a DVD-9) to maintain picture performance and single sided playback. The versatility provided by having a suite of discs available to the author/replicator is a DVD advantage.

Making A Complete Frame

Data is retrieved from the disc and is stored/assembled into a complete picture by RAM buffer in the player. Once a complete field is assembled, it is released by the buffer. The quantity of data being retrieved does not affect the speed at which a complex picture is assembled or processed.

DVD on CD

Sometimes referred to as cDVD, this method is an exciting  use of DVD technology that consists of burning  DVD-formatted volumes onto a CD instead of a DVD disc.  This is far less expensive than mastering to a DVD or DVD-R.  Thus, it’s an inexpensive way to deliver short projects using  DVD-quality content on an inexpensive CD.  The content used must meet all the DVD standards.

You can use MPEG-1 video at SIF resolution with a bit rate of about 1.856 Megabits/sec or full-motion, full-resolution MPEG-2 video not exceeding 8 Megabits/sec. The audio formats supported in the DVD specification include PCM, AC3 audio, or MPEG-1. These types of CDs can only be played back on computer systems equipped with the appropriate software DVD player. It’s a good idea to include a DVD software player on the CD to ensure your client will be able to view your cDVD.

On the systems we supply, once satisfied with your last edit, you export your project directly from your editing timeline in either MPEG-1 or MPEG-2, avoiding the quality loss associated with an extra tape generation as required in a conventional DVD authoring process. Therefore, the process remains totally digital to ensure the highest possible quality.

For MPEG-1 encoding, you can use the MPEG-1 encoder that’s included with the Adobe Premiere 6 release CD or any other standard MPEG-1 encoder. The MPEG video files can then be taken into your DVD authoring package such as the excellent Reel DVD by Daikin

In your DVD authoring package, you can then create menus to provide navigation and include audio and graphic elements for a complete DVD. The DVD is then output as a DVD volume that contains all the necessary formatting for a DVD.  This volume can then be written to a CD either directly from the DVD authoring package or using any standard CD writing software. The obvious limitation of this format is that your project can only be played back on a computer and not on a set-top DVD player.

Of course, in addition, the program length is limited to the 650-MB capacity of the CD.

Video CD

Video CD is the only CD format with a specific standard.  It’s similar to a DVD, but with fewer options for interactivity. The complete Video CD standard defines the navigational elements, audio and video compression, graphics content and total bit rate required. Video CD uses video and audio compressed in MPEG-1 with a total bit rate not exceeding 1.5 Megabits/sec. The video used is SIF-resolution MPEG-1 compressed to meet the "White Book" standard. The audio used is MPEG1 layer 2.  (LSX encoder provides preset values that conform to this standard)

Previously only playable on set-top Video CD players or on personal computers, the Video CD had only attained popularity in certain markets, such as in Asia, however, the advent of set-top DVD players that can play Video CD has revitalized the interest in Video CD because with the cost of DVD still high and the potential for good quality from a CD, the Video CD has become a viable alternative to CD and DVD.

All CVP NLE platforms provide the capability to export Video CD compatible MPEG-1 files using your preferred MPEG-1 encoder or the MegaPeg LE encoder included on the Adobe Premiere 5.1 release CD.  The audio and video elements then must be brought into a Video CD authoring package. Many CD-recording software packages include the capability of creating Video CD. Some  of the most popular applications include Adaptec Easy CD Creator, Win-on-CD, NTI CD Maker, and Nero. You should  verify that your CD-recording software is able to create Video  CDs. The CD's can then be written using standard CD recorders or sent for mastering at duplication facilities.