When home video taping first became available there were two format - VHS & Betamax, and VHS won. Now, ever since the arrival of camcorders and then digital video, there seems to be an ever increasing array of formats. Which should you use? This tutorial attempts to guide you through the maze.

Digital video formats are always a trade-off between size and quality. Maximum quality is only achieved by sampling at a high rate and not compressing the data. Doing this with broadcast-quality video requires about 20 MB/sec of digital data (or 1.2 GB/min). For most purposes this is an impractical amount of data, so various compression schemes are used to reduce the amount of data without noticeably reducing quality.

The DV format now universally adopted in consumer camcorders compresses the data to a fixed rate of about 3.5 MB/sec. This preserves visible quality very close to broadcast, and better than any consumer analog tape format. The fact that the video data is recorded on the tape digitally means that it can be copied by Firewire to a computer disk and back to tape, without any of the loss of quality associated with analog copying.

DV and Mini-DV are the same thing, Mini-DV just uses a physically smaller tape cartridge with shorter running times.

This format is almost exclusively used to capture analog video for digital editing. It's usually implemented by the hardware of a capture/replay card, and different manufacturers tend to implement it slightly differently. Hence, M-JPEG files captured by one card may not be playable on another. The compression, and thus resulting data-rate, is normally adjustable. To retain the quality of Hi8 or SHVS, a data rate of about 3 MB/sec or higher is required. 1 MB/sec is roughly equivalent to VHS quality.

DVDs are recorded using MPEG-2 compression. The MPEG compression schemes achieve much lower data rates than DV or M-JPEG for the same visual quality, but are more difficult to edit than DV or M-JPEG.

The VCD format uses MPEG-1 compression, at a data-rate identical to that of an audio CD. The quality is low, slightly lower than VHS, although a good player can sometimes make it appear slightly better. The compression parameters for an MPEG-1 file that can be used to make a VCD are very precisely defined. The current (2.0) standard for VCD also allows still images to be included, enabling such things as a CD photo-album to be created.

This is a rather loose standard that allows MPEG-2 format data to be used on a CD medium. Many current DVD players will also play SVCD-format CDs. It's a useful interim format for putting reasonable quality video on recordable CD, until recordable DVD is available at consumer prices. There is much more flexibility in the compression parameters of MPEG files written to SVCD, unlike VCD in which there is none.

The MPEG formats are not limited to CD and DVD use. Digital broadcast TV uses virtually the same MPEG-2 format as DVD. Both MPEG-1 and MPEG-2 formats can be used at a variety of frame-sizes, frame-rates, and data-rates. MPEG-1 files can be small enough to be sent over the Internet, especially if a small frame-size is used.

The main differences between MPEG-1 and MPEG-2 are that MPEG-1's maximum frame size is 1/4-frame and that MPEG-1 only allows one field/frame. MPEG-2 supports up to full-frame and two fields/frame, corresponding to TV. MPEG-2 also allows higher quality audio (the popular MP3 format for audio compression is in fact "borrowed" from the MPEG standard).

There are several other compression formats that can be used on a PC, including for example Intel's Indeo and Apple's Quicktime. These can be an appropriate choice for video that is targeted exclusively for viewing on a computer. Microsoft have also recently introduced their Windows Media Format, which is a deliberately low quality, low data-rate format designed specifically for Internet use.

However, none of these are usually relevant to work that involves any sort of external medium (tape, CD, DVD). The two formats that are becoming increasingly predominant in these areas to the exclusion of others are DV and MPEG.

There is a fundamental difference between the DV and MPEG styles of compression, which affects the editing process. DV (and M-JPEG) stores each frame as a complete image - it's the digital equivalent of a cine film. MPEG only stores some complete frames -- the remainder is stored as differences between adjacent frames. This enables MPEG to achieve its much more efficient compression, but means the editing process is more complex for a computer to handle.

The consequence is that simply making cut-edits on an MPEG file can result in quality loss, since frames have to be re-computed, and the more edits are made the greater the loss. It can also make the editor software sluggish to respond because of the amount of intense computing required. DV & M-JPEG, on the other hand, do not suffer from this problem because frames do not depend on each other. Images only have to be re-computed where transitions, special effects, etc. are used.

Where an MPEG result is required, the best quality is usually achieved by editing in DV or M-JPEG format, then rendering or converting the result to MPEG. Even better is to also save a DV copy as an archive in case subsequent changes are needed.

Devices are available that will capture directly to MPEG. They are appropriate if little or no editing is required. The main advantage is that they save the time required to generate the resulting MPEG file, which can be a slow process even on the latest fast processors.