Video  Video consists of image frames captured from real motion and shown in succession  Animation is similar except that the frames are synthesized  The.

Report
Video

Video consists of image frames captured from real motion and
shown in succession

Animation is similar except that the frames are synthesized

The frame rate is the number of frames shown per second (fps)



One minute of uncompressed video requires significant storage: 1.6
GB to 1.85 GB
Compression can be accomplished on the recording device or using
a video capture card
The IEEE 1394 (firewire) interface standard is often used for video
capture
1
Analog Video




An image on a TV is drawn row by row from left to right,
starting at the top left of the screen
After the image has been drawn, it moves back up and
begins again
The refresh rate of a TV is the number of times it draws a
new image per second
The refresh rate is measured in Hertz (basically, per
second)
2
Analog Video

There are two main TV formats:
• NTSC - National Television System Committee
• PAL – Phase Alternating Line



NTSC is used in America and Japan, while PAL is used
in Europe and Australia
An NTSC TV has a refresh rate of 60 Hz, while a PAL TV
has a refresh rate of 50 Hz
Standard movie frame rate: 24 fps
3
Video Codecs

An algorithm that shrinks the size of a movie to allow
the movie to be played on a computer or from a
network

codec is short for compressor / decompressor

Use lossy compression

Some video codecs are “software only” while others
require special hardware
4
Video Codecs

Most codecs are block-oriented

Divide the frame into blocks


Sometimes you can see the block of pixels when a video
plays
Codecs differ in how the blocks are encoded
5
Video Architectures

Handles and synchronizes video files

An architecture creates file formats with its codecs

Examples of Architectures:
•
•
•
•


QuickTime
RealSystem G3
Microsoft Windows Media
MPEG
The architecture you choose determines which codecs
you can use
Not all codecs are available for every architecture
6
Compression Techniques

Codecs compress video using two techniques:
• Temporal Compression
• Spatial Compression

Temporal compression eliminates information that is not

Spacial compression eliminates information that is not

necessary for visual continuity over time
necessary for continuity over area
Other compression can be accomplished by reducing
colors, frame rate, and audio quality
7
Temporal Compression




Temporal compression looks for differences between
consecutive frames
A keyframe is a frame that is used as the basis for
differences
Keyframes contain the entire image
A delta frame (or difference frame) contains only the
areas that are different from one frame to the next
keyframe, delta, delta, …, keyframe, delta, delta, ...

Several keyframes are chosen throughout the movie
whenever significant changes occur
8
Spacial Compression


Generally focuses on individual frames
If an area is all one color, the area is defined using
geometry and the color stored once

Combines bitmap and raster graphics concepts

Temporal and spacial compression work in concert
9
Asymmetric vs. Symmetric

A codec can be asymmetric:
• Encoding process is slower than the decoding process
• Better suited for movies that are prerecorded and played back

Or symmetric:
• Encoding process takes about the same time as the decoding
process
• Better suited for live broadcasts or video teleconferencing
10
Video Codecs

Some popular video codecs:
• Sorenson Video
• Cinepak
• MPEG:
– MPEG-1
– MPEG-2
– MPEG-4
• Real Video:
– RealVideo G2
– RealVideo 8
• H.261 / H.263
11
Sorenson Video Codec

Developed by Sorenson Vision, Inc (www.s-vision.com)
• High compression
• High-quality/low data rates
• High-end processors NOT needed

Recommended for:
• Web video (fast viewing rates)
• CD-ROM video (fit more movies on a CD-ROM)

It is asymmetric

A hardware specific version is also available
12
Sorenson Video Codec


Sorenson is supported by the QuickTime 3 and 4
architectures
It uses:
• variable encoding
• vector quantization
• motion compensation
13
Variable Encoding




Adjusts the bit depth according to how much action
the scene contains
Two-pass encoding technique
Movie is analyzed to determine “easy” and “difficult”
sections
Movie is compressed intelligently:
• Attempts to give each frame the optimum number of bytes
14
Vector Quantization





Image divided into segments (vectors)
The mean value of each segment is subtracted from the
individual pixels in the segment (residual vector pattern)
Each residual segment pattern is compared to a set of
previously determined patterns (stored in a codebook)
The codebook entry pattern closest to the residual vector
pattern is determined
The binary address of the codebook entry is sent to
receiver to be decoded
15
Vector Quantization



The decoder only needs the address of the residual
pattern and the mean value of the individual segments
If codebook pattern entries are determined ahead of time
according to test imagery, the quality of the movie will be
good
The number of codebook entries can be increased
substantially without too much effect on the bandwidth
needed
16
Motion Compensation




From frame-to-frame, there is usually some translation
of blocks of pixels due to motion
Successive frames in a motion sequence are searched
to determine the motion compensation vectors
Motion compensation vectors indicate the direction and
distance of translation of block of pixels that has moved
The whole frame does not have to be translated - the
vectors contain the information on how to reconstruct
the new frame from the previous frame
17
Sorenson Tools




Allows for a media key so that the video can be locked
with a password
Tries to determine when new scenes begin and marks
them as key frames (smart screen change detection)
Allows the user to insert a custom watermark on all
frames
Allows the user to specify intervals at which frames
should be dropped to increase playback performance
18
Cinepak

Codec maintained by Compression Techniques, Inc.

Medium-quality CD-ROM video

Does not require a lot of processing power

Is asymmetric

Uses vector quantization

Cinepak video codec is supported by the following
architectures:
• Video for Windows
• QuickTime
19
MPEG




Developed by Motion Picture Expert’s group
MPEG is both a file format and a codec (a file format that
employs compression algorithms)
Usually requires special hardware for encoding
Usually requires substantial CPU power/special
hardware for playback

Mainly used for high-end desktop video

MPEG is an open standard
20
MPEG

MPEG is a family of codecs containing:
• MPEG-1
• MPEG-2
• MPEG-4 (approved in 1999)

MPEG supported by the following architectures:
•
•
•
•
•
QuickTime 3
DirectShow 6
Video CD
DVD Video
Windows Media
21
RealVideo

RealVideo a family of codecs containing:
• RealVideo 8
• RealVideo G2

Usually used for streaming video

Both RealVideo codecs are scalable
• compensates if the connection bit rate is slower than the bit rate used
to encode the video
• Allows for smoother playback on a wide range of machines

Asymmetric

Requires a high-end computer for optimal encoding/decoding

RealPlayer 8 supports variable encoding
22

similar documents