Video Encoding - DVD


To create a DVD disc, the following specific formats must be followed.

MPEG II video elementary stream

  • Video : MPEG-II
  • Frame resolution maximum : 720 x 480 pixel (NTSC), 720 x 576 pixel (PAL)
  • Frame resolution : 702 x 480, 354 x 480 (NTSC), 704 x 566, 354 x 720 (PAL)
  • Frame rate : 29.97 fps (not 30 fps in NTSC), 25 fps (PAL)
  • Size of GOP : 36 fields/sec (NTSC), 30 fields/sec (PAL)
  • Sequence header information required within every 1.0 sec
  • Maximum bitrate : 9.8Mbps

Popular encoding for movie:

  • VBR rate : min 2 Mbps to max. 8 Mbps (ave. 3.5 Mbps for DVD-5 or ave. 5.0 Mbps for DVD-9)

Audio Format

PCM audio details

  • Linear PCM (pulse code modulation) is loss-less, uncompressed digital audio
  • Same format is used on CDs
  • DVD-Video supports sampling rates : 48 or 96 kHz with 16, 20, or 24 bits/sample
  • 1 to 8 channels in each track
  • Maximum PCM bitrate is 6,144 Mbps

To achieve high-fidelity sound reproduction, select 96-dB dynamic range of 16 bits or even the 120-dB range of 20 bits combined with a frequency response of up to 22,000 Hz from 48 kHz sampling.

Sample frequency : 48 or 96 kHz
Sample size : 16, 20, or 24 bits
Channels : 1, 2, 3, 4, 5, 6, 7, or 8

Dolby Digital audio details

Dolby Digital (AC-3) is a multichannel digital audio format, compressed using perceptual coding technology from original PCM with a sample rate of 48 kHz at up to 20 bits of precision.

  • Dolby Digital standard sampling rates : 32, 44.1, 48 kHz (only 48kHz allowed on DVD)
  • Frequency response :
    • 3 Hz to 20 kHz for the main five channels
    • 3 to 120 Hz for the low-frequency effects (LFE) channel
  • Encoding bitrate : 64 to 448 kbps
    • with 384 kbps being the normal rate for 5.1 channels
    • typical bitrate for stereo is 192 kbps
    • monophonic audio is usually at 96 kbps for music or 64 kbps for voice
  • Dolby Digital 5.1 : 1, 2, 3, 4 or 5 channels plus subwoober (.1) channel is option
  • Downmixing : all Dolby Digital decoders are required to perform a downmixing process to adapt 5.1 channels to 2 channels for stereo PCM and analog output
More about MPEG encoding

Taking original audio and video content (approx. 2.5GB of information per minute) and reducing the size of this data into a manageable stream of information that will be placed onto a DVD disc, is a complex task. Data about each picture frame, like contrast and colours, as well as the audio information, must all be reduced to a size of data that will fit on a standard DVD disc. In short, a movie of 2 hours would be about 150GB of uncompressed data that would need to be reduced to 4.7GB in order to fit onto a DVD disc.

MPEG encoding keeps the information about a reference frame, also known as an "I" frame, then scans subsequent frames (on a frame to frame basis) for differences in the content. The delta data is compressed with the reference frame data and delivers a smaller block of data.

Exhaustive motion estimation (full PEL)

In order to provide bit savings, frames in the encoder are compared to each other then the original baseline frame data is saved with the delta data. Most lower cost encoders provide a random search on the pixels of each macroblock seeking out the delta data. The gross search process will miss subtle details from frame to frame resulting in block artifacts in the output stream. This is called partial PEL.

Some encoders manufacturer go for full PEL. For example. Zapex designed its own VLSI motion estimation chip that can fully utilize its patented motion estimation algorithms running at well over 250 giga-operations per second. The Zapex encoder employs two very complex schemes: - First, a motion estimation technology that sets new standards for digital video encoded data. Rather than a gross estimation through random macroblock sampling, Zapex applies a full search, qualitative motion estimation to every macroblock, achieving razor-sharp details that eliminate block artifacts, "noise," and jerky motion. - Second, the Zapex encoder optimizes the output through macroblock-based Adaptive Field Frame encoding to determine which of multiple quantitative algorithms will produce the most professional quality at the lowest bit rate on every macroblock within the frame.

Manual entry point ("I" frame) insertion

A Group Of Pictures (GOP) in the MPEG signal is a sequence of "I" frames, "B" frames, and "P" frames. These contain compressed information, that together make up the entire picture - an entire GOP must be decoded to see the picture.

MP@ML, DVD compliant video output

Zapex provides MP@ML (Main Profile@Main Level) MPEG-2, which is the compliant video profile for DVD specifications, Please note that 422P@ML, which is used for creating editable MPEG streams, is not DVD compliant. Video output (full 720x480), half D1 (352x480), and SIF (352x240) resolution.

Program stream

Program Stream (PS) is necessary for users requiring MPEG encoding in multimedia, archiving, and store & forward network video applications.

Automatic scene change detection & adaptive GOP ("I" frame insertion)

A Group Of Pictures (GOP) in the MPEG signal is a sequence of "I" frames, and "P" frames. The GOP structure is defined by an "I" frame followed by a sequence of "B" and "P" frames ending with an "I" frame. Without automatic scene change detection, the encoder consistently utlizes the same GOP structure, despite changes in the content's complexity, resulting in a loss of quality.

When using manual entry point ("I" frame) insertion for scene change detection, subtleties within the picture frames are difficult to separate. Zapex's algorithms can verify the difference between frames and automatically detect where a new GOP ("I" frame) will be inserted. The end result is improved video quality.

  iSchoolMedia logo (2K)For any comments concerning this page, e-mail to info@ischoolmedia.com.

The material in this document is for information only and is subject to change without prior notice. Whilst reasonable efforts have been made in the preparation of this document to assure its accuracy, iSCHOOL NEW MEDIA LIMITED assumes no liability resulting from the use of the information contained herein. Company names and products are property of their respective owners.