Video Compression

by Vaibhav Mangrulkar

In this era of the information "superhighway," Pentiums, and PowerPCs, we demand that multimedia be just as fast, cheap, and powerful. Video playback, one of the prime components of the multimedia standard, was the bane of all soft ware developers just several years ago -- some videos required so much space and transfer time that translocation was impractical. Binary and text compression programs, such as the popular PKZIP, have been available for a long time; However, they cannot compress images and movies very well. This is because such programs are based on the idea of replacing repeating blocks--such as ABCD--with smaller blocks--such as E. This cannot be applied efficiently to video because there are very few such repeating b locks. With the introduction of reliable and effective video compression, videos can now be stored into much smaller files with very little detriment in quality. Moreover, the time required to transfer these videos from one computer to another has been gr eatly decreased.

Similar to other computer components, video compression software and hardware (codecs) are produced by many different companies, each having advantages and disadvantages. While software compression codecs have the advantage of low cost, they lack compress ion speed. With the powerful Pentium processors, this is no longer a problem for those who can afford the upgrade. The other option, hardware-based compression codecs, boasts VCR-like compression and decompression speed. In these codecs, special hardware works solely on video files, taking the workload off the CPU. The video accelerator card acts exactly like a graphics accelerator card and takes the SVGA graphics-loading work from the CPU. Therefore, hardware codecs earn the higher ranks in terms of imag e clarity and smooth flow. The disadvantage is that users must have that specific card for decompression, which costs about $300. On the other hand, software decompression codecs are freely available from the Internet.

Software developers who use video compression also consider the compression quality of various codecs. The underlying principle is to tighten the video file into the smallest possible size without sacrificing playback quality. One option, video compressio n via intraframe analysis, is done by algorithms that divide a video frame into blocks that are eight by eight pixels. Another method, interframe analysis, is where the algorithm explores between frames. Both use "lossy" compression, a method that searche s for and replaces redundant data in the files, in order to reduce the overall size of the file. Furthermore, these codecs can be classified as either symmetric, where the algorithm splits the time for compression and decompression equally, or asymmetric, where compression takes more time to reduce the file size even more. Once the dropping of redundant data takes place, the codec processes the remaining "useful" portions of the file and statistically forms images, which become encoded with an RLL (run-le ngth limited) scheme. MPEG, a very popular compression format among companies, uses discrete cosine transform (DCT) wave equations to compress the individual blocks. DCT reduces the file si ze by dropping the number representations of the finer pixels in the video. Therefore, while some fine detail is lost, it is not significant enough to compromise the playback quality. Yet another transformation and compression technique is done by RLE (ru n-length encoding), in which the codecs search the blocks at different angles to look for more frequently occurring strings.

Currently, the industry standards for video compression are Microsoft Video for Windows and Apple Quicktime. Video for Windows uses the antiquated AVI (storage and playback of consecutive graphics frames) format and is very common among video compression hardware and software developers. Quicktime's advantage lies in its compatibilty with both Windows and Macintosh environments.

Apple Quicktime Homepage
Microsoft Video for Windows
MPEG