R.B.4    1 A / V   W O M A N   & A V I E W 2 0 0 7   /   8

1 SERIAL NO. IM  448199

In the above table the channels are listed in approximate order of their average bitrate (given in kbit/s to the nearest 250). The greater the bitrate, the lesser the image compression and hence the better the picture. In simple terms, a lower bitrate is achieved by extending the GOP length or by reducing the reference i-frame's image quality as one might when compressing a JPEG. It's worthwhile knowing that 5000 kbit/s is the DVD-standard bitrate (maxing out at 8000). 1000 would be roughly equivalent to VHS while HD is 15000. 

It should be born in mind that the bitrates listed are an average and that the actual bitrate is variable: if a channel is showing something particularly graphically intense its bitrate will increase; if it is showing a black screen its bitrate will be tiny. Taking that information on board, BBC 1 can be seen to broadcast pretty much at DVD quality.

The coloured panels in the table show "statistical multiplexes", that is to say groups of 'bandwidth buddies'. These are channels which share / fight for an allocated section of bandwidth within their multiplex. I1, C4, C5 and B4 get to take two slices or so of cake in such battles.  B1 is now the only channel to not partake in such statistical multiplexing (apparently a legacy of its regional variations set-up and the low technology of Mux1) but it intends to tie up with its fellow Mux1 channels at the turn of 2009. The column labeled "SHARE" indicates the degree of bandwidth competition for each channel (and should also help give an idea as to the maximum bandwidth available): "1/4", for instance, indicates that the channel is one of a group of four such bandwidth buddies.

An interesting observation to make here concerns E4+1 and ITV2+1. Most significantly, both are broadcast at a higher resolution than their live counterparts, meaning that all other factors being equal the timeshift stream should provide the superior picture. The live channels do use lower maximum GOP lengths, but given that the live and timeshift channels have the same average bitrate, the i-frame compression of the live channel is likely to be greater. The usual caveats of course apply here: factoring in the temporal dimension, the longer GOPs of the +1 channels will present their own problems, but GOP lengths need not be at their maximum extent. Given busy movement in footage, the +1 channels can shorten their GOPs resulting in a picture compression factor equal to (or, if we wanted the same GOP length and bitrate, slightly greater than) that of the live channels. But still the +1s would have the advantage of a higher broadcast resolution. The reason for such an unusual discrepancy is simply an accident of history. The later slots (those taken by the +1s) just happened to be a bit roomier than the originals. It's perhaps worth pointing out that the bitrate difference between a 720 res i-frame over an 18 long GOP and a 544 res i-frame of equal compression over a 12 long GOP is only about 100kbit/s, so the higher resolution of E4+1 really is worth considering, especially for any episode of Friends where Chandler is comatose (and hence not causing the GOP maths too much difficulty with his flailing arms).