Piano Pre-Skills: Giving Your Kids a Musical Head Start

Shooting High Definition Video Footage in Italy

Psychic Palm Readings

Maximizing Your Home Theater With Minimal Space

Acting - Techniques in the 20th Century

A normal 625 line frame (we will use the PAL standard in this example)takes 1.2 Megabytes of data if no compression is used. As there are 25 frames per second, this will give us: 1.2 x 25 = 30 Megabytes per second

So a DVD with 4.7 Gigabyte of storage will give us: 4700 / 30 = 156.6 seconds or 2.6 minutes of video playback.

Obviously some way of reducing the amount of data that needs to be stored on the disk is needed.

One way to compress the data stream is to analyse each frame and remove information that is repeated within the frame. This frame by frame compression is called JPEG. This might be ideal for individual frames but is not sufficient for movies. Jpeg looks best with compressions of 5:1 or less, even this will only give 13 minutes of video on a DVD.

A better method is to use MPEG compression. This works by taking the first frame and compressing it as above. This frame is still quite large as it contains most of the information about the frame, this is called the I frame. The next frame of picture information is compared to this I frame and only the difference is stored. There are two types of difference frames, B and P frames, B frames are bi-directional, and P frames are predictive. The fact is that B and P frames can be quitr small depending on the amount of movement within the picture information. Typically with MPEG we can realise cpmpression factors of upto 30:1.

The grouping of I, B and P frames is called a GOP structure (group of pictures) the default stucture of a GOP within a DVD is usually IBBPBBPBBPBBPBBI, 4 P frames within each I frame sequence, and 2 B frames between each P frame sequence. A GOP sequence of 16 will give a very good compression ratio and take up less disc space but may present problems as below.

In the DVD authoring process certain functions have to happen on an I frame and as each GOP can only have one I frame then for some videos a smaller GOP structure is more flexible but take up more space. Foe example a chapter point has to happen on an I frame, if the GOP structure was 16 frames you might find the nearest I frame upto 8 frames away which might result in an inaccurate position to begin a new chapter.