One concept I found particularly interesting was the idea of inter-frame compression. The idea behind this type of compression is that you don't want to transmit pixels that aren't changing from frame to frame because that would be a waste of bandwidth. I realize though that those are a lot of words/ideas some of you may not be familiar with so let's take it from the top.
Pixels
Every image is made up of several tiny colored dots that will form that image to what you and I see. The more pixels the image has, the more realistic it will look.
Take for example the two images above. The one on the right has a much higher pixel density meaning it has more pixels per inch. Obviously the quality on the image on the right is far superior.
Motion Pictures
I bring up the idea of motion pictures because I mentioned the process of frame to frame compression. That's all motion pictures really are. They are separate frames of one image per frame that, when compiled together at extremely fast speeds, look as though there is motion. Inter-Frame Compression
This brings me to the idea of Inter-Frame Compression. The first thing to know is that a key frame will be identified. This is the first frame in the series that acts as the base point for comparison. It is the initial frame. In the image below, the first image all the way to the left is the key frame. The I Frames are the next frames in the series of images. So let's say this is a small sample of frames in a video that you want to compress. You could compress each of these images individually but this is an unnecessary use of bandwidth. This does not take advantage of the capabilities of video compression. You'll notice that, from frame to frame, some of the images do not change. These pixels then will not be transmitted when compressing the video. Think of a lecture that you've seen where the teacher is hardly moving around the classroom. So many of the pixels are identical from frame to frame, so why waste space transmitting those identical pixels. There is just a command that will copy that part of the identical pixels. For example, in the image I created above, The second image is identical, so instead of the second frame being transmitted, it will simply be copied. Moving on to the third frame, once again the identical pixels will be copied, and the new pixels that make up the red line will be transmitted.
Drawback
1) Data Corruption - If the first frame is corrupted then the corruption will carry on until a new key frame is established.
Acknowledgments: http://www.gadgetell.com/tech/author/Gregadeaux/ © 2011 Gadgetell LLC. See our Privacy Policy.