posted on Jun, 27 2010 @ 01:24 PM
Originally posted by plopunisher
I am a complete video noob so would someone give me one legitimate reason as to how this could be attributed to compression of the original video
stream?
There's a lot of different parts of video compression. But one important one is "don't transmit parts of images that haven't really changed".
This is what we call "intraframe compression", as opposed to taking out the more redundant bits of a single frame.
In intraframe compression, the sender provides a basic description frame that contains everything, generally called an i-frame. An i-frame is like a
snapshot. From there, the compression engine will look at the next frames coming in, and try to find parts that haven't changed, and just send the
chunks that have, in what's called a p-frame or b-frame, which one gets sent depending on what's going on in the picture and how good the
compression engine is.
In images where the engine is going for maximum compression, and maybe it's not that good to begin with, it may have trouble separating out the parts
of the picture that are changing from the parts that are static, especially if the moving and static parts look a lot alike, like guys in camo in a
pretty well matched low-res background, as you see in the video. When that happens, you get things that look like what you see until the compression
engine catches up and sends an i-frame to correct its mistake. It looks just like this.
If you notice the vehicles in the background, I believe you'll see one of them has a big chunk missing at one point, same thing.
edit to add: That i-frame/p-frame thing is why, when you watch digital video from your cable company, directv or dish network, when you see a
lightning flash or camera strobe go off, the entire picture breaks up into pixellation for a second. The "flash" looks like a complete picture
change to the compression engine. It therefore sends a full i-frame at the beginning of the flash, and another at the end. But there's not enough
bandwidth in the channel to do this, so what happens is it sends as much as it can - the intraframe compression (that's the one that works on single
frames) then has to cope with not having enough data to make a good image, and you can see the compression blocks until it catches up. That doesn't
happen on well-authored DVDs because the authoring company takes time around the flashes to send 'catchup' b-frames before the flash, or they encode
the blink as a moving object instead of a background image change. The DTV/cable people can't do that in real time.
[edit on 27-6-2010 by Bedlam]