Joined
·
4,146 Posts
You are correct, grasshopper! Very astute observation.jdspencer said:
MPEG uses a "group of pictures" where certain sorts of frames in the GOP (B and P frames) are dependent on other frames (I frames) to be sent before decoding. This is primarily what accounts for the latency, waiting for successive frames before decoding them into video frames.
When you tune to a channel, typically in the middle of the GOP, you have to wait for the GOP to end (and an I frame to be sent) before you get acquisition, because the P and B frames sent before that I frame have no reference and can't be decoded. During the acquisition delay, the video is muted to black.
The GOP for MPEG-2 (as used by ATSC and DVB HD) means an I-frame will come along every half-second or so. For MPEG-4, I frames can be as far as 200 frames apart (6 seconds or more) although they are typically sent by DTV more often to prevent having to wait that long for acquisition. Camera shot changes are also sometimes accompanied by an I frame to prevent pixellation. This is the very same reason FFWD is smoother with MPEG-2 than MPEG-4.
