[time-nuts] time-nuts Digest, Vol 114, Issue 3
Magnus Danielson
magnus at rubidium.dyndns.org
Sat Jan 4 02:30:10 UTC 2014
On 02/01/14 16:27, BD Systems Inc. wrote:
> Re: Measuring TV delays
>
> There's more delay involved in a live feed than most people assume. Let's start with the live feed or remote. Most live feeds with live crowds will use a 9 - 10 sec delay to avoid FCC issues with language.
Only applies to the US then.
> Then it's usually one satellite hop (1/4 sec) to get to the broadcast control center.
Actually, fibre-based links is becoming widely used now. There are many
players in that field now, including sat-operators.
> Frame syncs usually add a 1 or 2 frame delay.
Frame-stores is needed to convert arbitrary video-phase into the
production house phase, such that line-stores in mixers can be used.
Frame-stores can also provide Time Base Correction, i.e. changing the
frame-rate of the signal. This is done by either dropping a frame or
duplicating a frame. This typically involves sample-rate conversion of
the audio too. How to handle time-information and other data while doing
TBC is... eh... quite unspecified. So don't expect it to work very well.
A frame-store adds 1-2 frames. There are examples where you regularly
run through 3-4 frame-stores as the signal steps through a setup.
> Distribution to local stations is usually done with another sat hop.
Fibre-based distribution is also quite common now.
In distribution you have usually compressed it into MPEG-2 or MPEG-4,
fit it into MPEG-2 Transport Stream (which is just a generic transport
format). Many mux and demux stages can exist, with restamping of
timestamps to hide the time variations.
> The local station does a pass through (to enable their local ads) and if you get it off-air then you're done. If you get their signal via cable (Comcast) or DTH (DirecTV) there is at least one more sat hop to go. If your set top box (IRD) has delay capability (TIVO), add another 1/2 sec.
Local insertion is usually very well matched time-wise as they are
automated.
Cheers,
Magnus
More information about the Time-nuts_lists.febo.com
mailing list