How does compression work, and how does it affect Picture Quality?

Echostar Knowledge Base
The Dish Network User's Resource


MPEG2 is a temporal compression. In other words, it takes the first frame of video, looks at the second, compares, and transmits only the changes in the picture, it then looks at the third frame, compares it to the second and so on.

MPEG2 lets you control how much of a change is required before it includes that update (more information, more bandwidth). It also lets you control how often you send a full picture (usually called a key frame).

It's this temporal compression that causes 'breathing' in the picture. This is when you will see a background image slowly lose coherence and suddenly snap back into focus.

There are other factors involved, like how much digital compression was used in the originating signal and how much action is in a scene. But most of the quality issues come from the amount of compression DISH uses. Remember, they're squeezing 10 or 12 channels on a transponder that carries one analog channel. If there's a lot of action on HBOE, DISH has to take bandwidth from another channel. If there's a lot of action on all of the channels, the bandwidth goes down for them all.

DVDs, on the other hand, require much less compression and can have their compression matched to each frame of video. DISH has to compress on the fly and cannot adjust compression for all of the channels on a transponder as accurately.

An excellent source of information on the subject of DBS compression can be found at http://www.hei.ca/mpeg2.html


What does compression have to do with picture “softness?”

They can “filter” the signal before they send them to the MPEG encoders to limit the amount of picture information -- I suspect DN at least some kind of analog “filtering” and cleanup before sending the image to the encoders -- a snowy OTA signal could easily require more bandwidth to “reproduce” than a perfectly clean signal with higher resolution due to the way MPEG works.

Filtering off the high frequency information will lower the resolution (horizontal resolution) and help the MPEG encoders keep the artifacts to a minimum -- when the encoders run out of bandwidth and they can't transmit what they need to, they stop sending individual pixels -- they group small blocks and make them the same color, brightness, etc. That's why explosions or other high bandwidth scenes (sports) often look “grainy” or I love this one, “digital”.