By Brian Boyko
Editor, Network Performance Daily

I’ve been getting better and better with video as I’ve worked here – and I noticed that many people don’t really understand the nature of digital video files – what makes one file big, what makes one small, what makes one 20 MB file look good, while another 20 MB file looks lousy. 

Considering that Cisco is claiming that video will take up 90% of all traffic by 2013, it might be important go over some of the ideas about what makes up digital video files at the low level so that you can understand how they impact your network. 

If you already know this, I apologize. 

Digital video files are, essentially, pictures.  Multiple pictures played in sequence, with an audio track.  And like digital pictures, they can be compressed.  Uncompressed video is like a photo in BMP format – perfectly accurate, but huge.  Compressed video is like a JPG file – it creates artifacts, but is much smaller. 

There are many compression schemes, with advantages and disadvantages, based on different ways to compress the file.  Some are more effective at reproducing more accurate information with smaller filesizes, but ultimately, from a networking perspective, it’s really not important which you’re using.  What is important is the resolution, framerate, and bitrate. 

The resolution is the size of the video in terms of how many pixels you can see on screen.  It’s usually in varying rectangle sizes, but it’s usually expressed as the size of the rectangle (i.e., 640x480) rather than the absolute number of pixels.   High definition television comes in two types: 1920x1080 (a.k.a. “1080”) and 1280x720 (a.k.a. “720”), and “720” is the format used for YouTube high definition video.  Standard definition television comes as either 720x480 (NTSC) or 720x576 (PAL).  

The framerate is the number of frames – or pictures – that are shown each second.  In a filmed Hollywood movie, it’s typically 24 frames per second (or 24p).  On television, and with most recording equipment, they usually show 30 frames per second in the U.S. and 25 frames per second elsewhere in the world – though, in order to get faster motion, they often stagger every other line of resolution and place them between every other frame – so when you watch 30 frames per second of TV, you’re really watching 60 “half-frames” per second - also known as 60 interlaced frames  (or “60i”).

Finally, the bitrate is the number of bits of information that are in the file for each second of video.  A file with a bitrate of 512kbps has exactly 512 kilobits per second in order to convey all the information it can about the video.  With more bits, you can contain more information. 

Ultimately, the only thing impacting the network is the bitrate.  Bitrate determines filesize for downloadable video, and bitrate determines bandwidth requirements for streaming video.   A 3Mbps SD video will be the same filesize as a 3Mbps HD video, for example, provided that they are the same length. 

The quality of the video, however, is impacted by resolution and framerate as well as bitrate.  

In one second of 640x480 video at 30 frames a second, and 24 bits per pixel, you have 210 megabits worth of information.  Representing that, with, say 3 megabits is a daunting task.  On the other hand, if you increase the framerate to high definition – 1280x720 – you get 632 megabits of data per second – and representing that with only 3 megabits is an even more daunting task. 

The higher the bitrate, the higher the quality of the video; but the higher the resolution and framerate, the more you have to increase the bitrate in order to get the same level of quality.  High definition files tend to be larger than standard definition files because they’re usually – but not always – rendered at a higher bitrate. 

This was one of the reasons that when NBC was covering the Olympics, they often sent the raw footage from Beijing in low-resolution, low-bitrate formats, and only sent high-resolution files to their editors in the U.S. when they knew exactly which shots they wanted to include in their broadcasts.

By way of metaphor, think of bitrate as a fixed amount of butter that you have to butter bread.  You could spread out a single pat of butter on a whole loaf of bread, but it wouldn’t taste very buttery.  Or you could just butter one slice with that pat, but the bread wouldn’t be very filling.

Where different compression schemes come in is that they’re ways to use the limited amount of bits in each second of video to represent all the information – some compression schemes simply look better than others.  Right now, the leader seems to be H.264, which is about twice as efficient as the file system used to store information on a DVD (MPEG-2.)  A 3Mbps H.264 file and 3Mbps MPEG-2 file will have the same filesize, but the H.264 file will simply look better. 

Bitrate is also important because for streaming video, the bitrate of the video is equal to the amount of bandwidth required in order to show a video without pre-buffering or jitter – especially important for live applications like teleconferencing.  Lowering the bitrate decreases the quality, but image fidelity is often less important than immediate response time in teleconferencing applications.  (There are other issues that can cause jitter even with sufficient bandwidth, so it pays to monitor your communications network.)

At any rate, I hope you find this information useful when dealing with bandwidth.  At least now, when someone complains about how long it takes to download a video file, instead of spending tons of time and energy on network upgrades, you might want to ask if you can get a lower bitrate file instead.