[ Previous | Next | Contents | Glossary | Home | Search ]
Ultimedia Services Version 2 for AIX: Programmer's Guide and Reference

Appendix B. Comparison of Video Codecs

A video codec generally introduces some distortion in the reconstructed output of the decompressor, relative to the video at the input to the compressor. In contrast, a file- or text-compression algorithm must reconstruct a file exactly from its compressed representation. The first compression algorithm is lossy and the second is lossless.

A file compressor may compress ASCII text by up to 2:1-3:1 but might not work with image data at all. A lossy compressor for still images may compress its input by 10:1 with no visible distortion in the decompressed output. Compressors for digital images or video can attain compression ratios of 30:1 or more and still produce quality reconstructed output. A good algorithm allows the user to specify the tradeoff between the compression ratio and the quality of the decompressed output. This is important, as decompressed video can use a tremendous amount of disk storage and transmission bandwidth, and different applications have different quality requirements.

Video codec algorithms tend to be complex, implying a tradeoff between obtaining good visual quality of the decompressed output (more computation) and a frame rate that is high enough to make the motion appear smooth (less computation). The computation can be distributed between the compressor and the decompressor in a particular codec. An asymmetric codec has a very simple decompressor, and a relatively complex compressor. A symmetric codec distributes the computation relatively equally between the compressor and decompressor. Asymmetric codecs are used to decompress and display video in software on computers of limited power where real-time compression is not important. Symmetric codecs are used where fast compression is important.

Ultimedia Services supports JPEG, Ultimotion Matinee, RTV 2.0/2.1, Indeo 2.1, Indeo 3.2, MPEG-1, H.261, and H.263 video. These codecs are intended for continuous-tone inputs; for example, digitized photographs and motion video. They are not intended for the compression of images of text or of halftone images.

To learn more about Ultimedia Services video compression algorithms, see:

The following table highlights the basic characteristics of each codec.

Algorithm Advantages
JPEG Still images or motion video

International standard codec for still images

High visual quality for still images

Relatively symmetric codec
Ultimotion Matinee Presentation-quality video

IBM proprietary

Very fast decoding in software

Asymmetric codec; slow encoder

Compresses to CD-ROM rates
RTV 2.0/2.1 and Presentation-quality video
Indeo 2.1 video Intel proprietary

Symmetric codec
Indeo 3.2 video Better video quality than Indeo 2.1

Intel proprietary

Very fast decoding in software

Asymmetric codec; slow encoder

Compresses to CD-ROM rates
MPEG-1 Excellent video quality

International standard

Real-time decoding on PowerPC 604

Asymmetric codec; slow encoder

Compresses to CD-ROM rates
H.261 Good to very good quality video

International standard teleconferencing codec

Real-time decoding on PowerPC 604

Real-time encoding on PowerPC 604

Bit rates from 40 kbps to 2 Mbps
H.263 Good to very good quality video

International standard teleconferencing codec

Real-time decoding on PowerPC 604

Real-time encoding on PowerPC 604

Operates at lower bit rates than H.261

For introductory information, see Comparison of Video Codecs.


[ Previous | Next | Contents | Glossary | Home | Search ]