Intraframe Editing Codecs
The de facto standard for video editing used to be Motion JPEG, or M-JPEG for short. As the name suggests, this is essentially a string of frames compressed as JPEG still images using a Discrete Cosine Transform (DCT) based on blocks of 8 x 8 pixels. As with JPEG still images, the video data rate is reduced by sampling the colour in a 4:2:2 format. This means that for every four brightness samples there are only two colour samples. This already reduces the data rate by a third, before any more sophisticated compression is applied.
No explicit frame size is specified for M-JPEG. For editing, formats usually operate at the standard SD resolution of 720 x 576 (which uses oblong pixels to create a 4:3 picture), or with square pixels for 768 x 576 (both in European PAL). The amount of compression can be varied from lossless (around 1.8:1) to high ratios, but quality drops considerably with compression above around 8:1.
The DV format used by digital camcorders isn't a million miles away from M-JPEG. The same DCT compression scheme is applied. However, there are some key differences. First, DV always uses a standard frame size of 720 x 576 pixels for PAL. The pixel aspect ratio can be varied to turn this from 4:3 to a widescreen 16:9 signal. The video signal is also sampled at 4:1:1 or 4:2:0, depending on the DV subtype, so for every four brightness samples there's only one on each colour channel (for 4:1:1) or two taken from alternate channels (for 4:2:0). DV also has a standard data rate set at 25Mbits/sec, which amounts to around 5:1 compression.
So DV requires over 3MB/sec even with SD. Considering that Full HD at 1,920 x 1,080 has five times the pixels of SD, DV clearly wasn't going to carry video into the next generation, as 15MB for every second of video would bring PCs to their knees. Even 3MB/sec was beyond the capabilities of the first DVD drives, so DV wasn't going to be much good for optical storage either. This is where interframe encoding has come in - particularly MPEG. Next week, we explain this complicated format in all its varieties.