What is MPEG video compression, and why should I care? With the rise in popularity of multimedia, more and more people are wanting to use video in their computer applications and presentations. Video can be a powerful and useful way to communicate an idea, demonstrate a product in actual use, include customer testimonials in a presentation, or provide training while on the road. Unfortunately, uncompressed video data is extremely large and can put an impossible strain on consumer PC systems. MPEG video compression is a method of reducing the size a video data so that it can be stored, managed, and played back on a computer. MPEG is used today in CD-ROM multimedia applications, Video On Demand television broadcasting, point to point satellite transmission of television, video archiving, mini-dish satellite broadcast, and VideoCD. Companies are distributing feature films, interactive training, and interactive product information on CD-ROM's using MPEG video. Anywhere video quality and interactivity are important is a good place to use MPEG video compression. The MPEG video compression standard was developed by the Moving Picture Experts Group to provide full-frame, full-motion video playback with audio, even at low bitrates. Video quality had to rival that seen on consumer VCR's, while maintaining bitrates that could be stored and played back on consumer CD-ROM's. The digital audio in MPEG is CD quality, stereo audio. MPEG is more ideally suited for motion picture playback at low data rates than any other video compression available (such as JPEG, Cinepak, Indeo, Quicktime, PLV). MPEG makes use of interframe encoding, algorithms that take into account redundant information between frames to more efficiently compress them. For this reason, MPEG is specifically geared toward moving pictures, not stills. MPEG takes much more computational power to create than it does to play back. Therefore, MPEG can be played on relatively inexpensive systems, despite it requiring special, more expensive hardware to create it. There are three different types of frames in an MPEG sequence. They are I frames (intra-coded, or self-sufficient), P frames (forward predicted), and B frames (bi-directional). I frames are the largest frames, because they contain all of the data necessary to reconstruct themselves. The compression in I frames is much like JPEG (still frame) compression. P are derived from I frames, and B frames are derived from both I and P frames. Therefore, P and B frames are smaller, but dependent on the rest of the sequence to play back. What are the different standards of MPEG? MPEG-1 was approved as an ISO standard in late 1992 for lower bitrate, 1/4 screen captures (352x240, doubled horizontally and vertically in playback), and frame based encoding. MPEG-1 is suitable for multimedia work which needs to be stored and played back on 1x CD-ROM's, or other limited bandwidth media. About 72 minutes of MPEG-1 (1.2Mb/s) can be stored on a 650MB CD-ROM. MPEG-2 is a new MPEG standard, and looks to be the standard the broadcast industry will embrace for digital video compression. MPEG-2 builds on much of the technology found in MPEG-1, and most MPEG-2 decoders can play MPEG-1 video. MPEG-2 captures are full-frame (as opposed to 1/4 frame captures in MPEG-1), and playback is 60-fields/sec. Because of this increase in resolution, MPEG-2 is optimal between 5 and 15Mb/s. Today, MPEG-2 is used in the broadcast industry for station-to-station satellite transmission of video programs, where perfect image quality is required and satellite bandwidth is expensive. About 17 minutes of MPEG-2 (5.0Mb/s) can be stored on a 650MB CD-ROM, and a 5x CD-ROM drive would be needed to play it back. The high bitrates and increased storage required for MPEG-2 currently make it impractical for use in CD-ROM-based computer applications. IBM resolves this MPEG-2 dilemma by using a hybrid format of MPEG-2 called MPEG-2 at half height resolution (MPEG-2 HHR). IBM provides the capability to play MPEG-2 HHR on the MPEG equipped Thinkpad notebook computer. MPEG-2 HHR has the ability to perform at lower bitrates (3Mb/s-4Mb/s) by using 1/2 frame captures instead of the full-frame captures found in standard MPEG-2. About 28 minutes of MPEG-2 HHR (3.0Mb/s) can be stored on a 650MB CD-ROM Drive, and a 4x CD-ROM drive would be needed to play it back. MPEG-2 HHR delivers the highest quality video available for a 4x CD-ROM. Some manufacturers are claiming to provide "editable" MPEG. This MPEG is not standard MPEG-1, but "I-frame only" MPEG. This format is made up of individually compressed frames (I frames), much like JPEG. The frames are large, so the bitrates of these streams are necessarily large. The user edits the I frame only MPEG, then is provided with an option to convert it to standard (I-P-B frame) MPEG. Once this conversion takes place, you have standard, non-editable MPEG. How is MPEG played back? MPEG video compression can be played back on three different types of systems: hardware decoders, software decoders, and hardware-assisted decoders. Hardware decoders are plug-in boards for PC's or MPEG chips built onto the computer's motherboard. Hardware decoders are generally the best option with full-frame, full-motion playback. The MPEG-enabled IBM Thinkpad has the advantage of a built-in hardware MPEG decoder chip. Software decoders range from freeware utilities to commercial applications, with varying features and performance. In general, software decoders tend to offer less than full-motion and less than full-frame playback. However, with faster CPU's and better algorithms, some decoders on the market today are achieving acceptable frame rates of full-frame video. Unfortunately, most software decoders use all of the PC's processor at playback, rendering the CPU useless for other tasks. Hardware-assisted decoders are hybrids of the above two options, where software decoders are paired with inexpensive MPEG chips to help them achieve better performance in playback. How do I get the best picture quality with MPEG? Image quality is often of differing importance in different applications of MPEG video compression. It often goes hand in hand with the production values of the original video and the professional quality of the MPEG encoding hardware and software. Feature films and entertainment titles are often produced to a much higher standard than instructional or industrial titles. The MPEG will never look BETTER than the image quality of the original video tape, so care in producing the best looking video tape and recording it on the highest quality format tape will always produce better results than the alternative. Use the highest bitrate your system can handle. The higher the bitrate, the better the image quality. The problem with higher bitrates is that there is always a limit to how much MPEG you can fit onto your storage media. There is also a limit in bandwidth to each system, and the limit of the playback card/software. For instance, the practical maximum bitrate of the MPEG-enabled IBM Thinkpad is 4Mb/s. Use the highest MPEG standard supported by your playback system. MPEG-2 delivers better picture quality than MPEG-1. Make sure that your bitrate requirements match up with your choice of MPEG standard. MPEG-2 at a very low bitrate (1.5Mb/s) is not better than MPEG-1 at the same bitrate. Both standards are optimized for a certain bitrate range. Use a compression service, such as IBM Interactive Media. Service bureaus typically have knowledgeable staff and high-end MPEG compression and video equipment to provide you with affordable, high quality MPEG video. IBM Interactive Media staffed their world-class facility with expert film and video professionals and digital media specialists. The facility is equipped with supercomputer encoders and professional quality digital video tape decks. IBM Interactive Media has become a leader in the MPEG compression business. What are the different formats of video tape can I use for MPEG? MPEG encoders require video for input, and any video source can be used, depending on what input video interface the MPEG encoder (or service provider) is equipped with. However, all video formats are not created equal. It is recommended that the highest quality video format be used, depending on what the budget will allow. Currently, there are four main video formats on the market today: digital component, digital composite, analog component, and analog composite. Digital component video (D1, D5, Digital Betacam) is the highest quality video format, because 3 component parts of the video signal are stored separately on the tape in a digital format. Since this separation exists, the signals have no way of diluting each other. This format has the ability to be reproduced with virtually no loss in quality. Digital composite video (D2, D3) is often used for mastering of video content when Digital component is not available or too expensive. The separate parts of the video signal are combined and digitally stored on video tape. Digital composite can be copied with little or no loss in quality, but because of its composite nature, it is inferior to digital component. Analog component video (Betacam, BetacamSP, MII, SVHS, Hi-8) is used as both a source and mastering media, because of its high quality, low cost, and small size. Analog component is just like digital component, except the separate signals are recorded onto the tape in an analog format. Because of this, duplication of the video tape will cause loss of the video signal between generations. Also be aware that not all analog component video formats are of similar quality. Betacam, BetacamSP, and MII are professional, high quality formats, whereas SVHS and Hi-8 are mostly used in the consumer market. Analog composite video (1", 3/4" U-matic, VHS) combines the separate parts of the video signal into one, and records it onto tape in an analog format. The larger the tape's bandwidth (partly a product of the tape's width), the better the signal. 1" and 3/4" are professional analog composite video formats, and VHS is a consumer standard. Notes on NTSC and PAL In order to understand the differences in videotape recording formats and how that relates to MPEG, it is necessary to understand a little of the history of television in the United States. From an economic standpoint, black-and-white television became viable well before color TV. When technology had advanced to the point that color televisions could be marketed at a relatively affordable price, the Federal Communications Commission stepped in. The FCC ruled that no color standard could be adopted which would render unusable the millions of black-and-white sets already in American homes. The National Television Standards Committee was formed to solve this problem, and the standard for color video which it developed for the U.S. became known as NTSC. To oversimplify a bit, the NTSC color signal overlays the black-and-white signal (the luminance component) with the color signal (the chrominance component) to create a composite signal. This signal has the advantage of being viewable on a black-and-white or a color monitor; it has the disadvantage of sacrificing a certain amount of quality to combine the components. The European television standard, PAL (Phase Alternating Line), works in a similar fashion. PAL has 625 lines of resolution, compared with 525 lines in NTSC, which provides a somewhat sharper picture; PAL also has a different frame rate - 25 frames per second compared with approximately 30 frames per second for NTSC. The differences are due to the differences in electrical standards between Europe and the U.S. Numerous videotape formats were developed to record the television signal. An exhaustive examination of them all would be book-length; for practical purposes, we can look at three formats which are heavily utilized.. D1 is one of the highest quality videotape recording formats currently available. This is due to the nature of D1- an 8-bit, digital component recording format. As you may have assumed, a component format is one in which the chrominance and luminance signals are recorded separately and only combined on the monitor during viewing. The term "digital" means that this component signal is recorded as digital (binary) information - a stream of zeroes and ones. What this means to the viewer is that the color, brightness, contrast, and sharpness of the picture are going to be pristine, and the recording of that signal is going to be exact - even through multiple generations (re-recordings of the same signal). In MPEG terms, this means that the compressed files will contain noticeable pixellation under the most demanding of circumstances - three-dimensional camera movement with subject movement in opposite direction and intricate, chaotic background, for example. Although D1 is one of the highest quality formats available, its cost limits its practical uses. The format which has become the accepted "broadcast standard", the highest quality currently in general use, is BetacamSP. BetacamSP is a component analog format. As with D1, the chrominance and luminance signals are recorded separately, but unlike D1, the signal recorded is an analog one. That is, the signal varies infinitely in amplitude and frequency. The signal is not quite as pristine, and the reproduction not quite as accurate, as D1. Thus, files compressed from this format show pixellation in slightly less demanding cases. Ironically, the lowest quality format is probably the best known - the immensely popular home video format, VHS. VHS is a composite analog format - it records analog signals, luminance and chrominance together. In addition, the recording heads used in VHS are not nearly as high-quality as those used in professional formats. Thus, the initial picture is much fuzzier (lacking in detail), and the color saturation much worse (colors can appear washed-out or overly bright). The basic flaws in the VHS format appear in MPEG as pixellation in virtually any background and total deterioration of the signal under any "stress factors" - fast motion, complex background, etc. A final factor when considering formats for compression is the generation factor. Each time one makes a copy of a videotape, it is referred to as a generation. Thus, if we shoot a video original (first generation), edit that (second generation), then make a copy of the edited master, we are actually working with the third generation. In digital formats, this is no problem at all - essentially, we could have an infinite number of clones (the digital term for copy) and each would be an exact duplicate of the original. With component analog, this is not a severe problem - BetacamSP can withstand several more generations, or dubs, before showing a drop in quality. With VHS, third generation quality is not usually acceptable for professional applications. Since the original edit master is, itself, the second generation, and it is usually unavailable for compression, the generation question becomes an important one. The essential thing to remember about MPEG compression is that the best it can do is preserve the virtues of the original. Unfortunately, MPEG also preserves, and usually amplifies, the flaws as well. With that in mind, it is best to begin the compression process with the highest quality videotape format available. BetacamSP or any digital format are the formats of choice for MPEG compression. Generally speaking, VHS or other "home" formats are not recommended for compression.