I'm wondering if anyone knows how this video is filtered and compressed:

If you watch it very closely, it looks like a still picture with a video superimposed on it.

The only movement seems to be the men walking around and the chips flying out of the chute. However, the machine seems to remain perfectly still while large branches are being run through it and the trees and leaves in the background don't seem to move either.

I suspect there is some kind of filtering that ignores changes in small numbers of pixels (such as slight movements of the machine and trees in the background), however changes in large numbers of pixels (such as the men walking around) are recorded.

I've never used any video editing features on my phone or camera, however I'm guessing there may be some way to adjust the filtering/compression for small VS large movements.

 

Modern MPEG compression techniques do look at differences between video frames and basically only send the difference between the frames
This, of course, significantly reduces the data that needs to be sent and the average video bandwidth required.
Don't think it actually tries to determine whether on not part of the image is stationary.

 

It's a very interesting and complex process. If you've every handled the bulk transmission and storage of video data from satellite transponders it's something you learn about quickly.
https://www.ustream.tv/blog/streaming-video-tips/keyframes-interframe-video-compression/
https://vimeo.com/blog/post/keyframes-inter-frame-and-live-video-compression

 

Modern MPEG compression techniques do look at differences between video frames and basically only send the difference between the frames
This, of course, significantly reduces the data that needs to be sent and the average video bandwidth required.
Don't think it actually tries to determine whether on not part of the image is stationary.

That's what I was thinking.

We had an early version of video conferencing (a 1997 Tandberg system) at our office and I noticed that the image would "pixelate" when moving the camera from one person to another, but it would be clear when the camera was stationary. I suspected it was transmitting only the differences in frames.

It's sort of like how the old fashioned animated cartoons were hand drawn by just changing the position of the characters while the background remained the same.

 

That particular video used the MPEG2 H.262 codec, also known as AVC1. It's possible that Youtube imposed this on whatever original format the user supplied.

And yes, most of these methods now look across multiple frames to save storage space by inter-frame compression. That's why they are not suitable for editing and generally have to be converted to an editing-friendly, frame-by-frame format before you can manipulate the footage.