[linux-dvb] Re: video pts differences on dvb-t in the uk

[Måns Rullgård <mru@mru.ath.cx>]

Torgeir Veimo <torgeir@pobox.com> writes:

> On Mon, 2004-10-11 at 19:08 +0200, Holger Waechtler wrote:
>
>> >>>http://212.227.36.83/linuxtv//index.php/The_DVB_Decoder_Challenge for a
>> >>>discussion of the topic.
>> >
>> >I thought the PCR/STC handling is to avoid clock / pts drifts, but I am
>> >seeing a constant, predictable offset. Does PCR/STC bring any such thing
>> >into the picture?
>> 
>> Audio and Video Streams contain Presentation Time Stamps (PTS) which 
>> reference to the PCR.
>
> Do you have any reference which discusses this with examples of the
> actual values involved?

ISO 13818-1 is a good starting point.  The PCR field contains the
value of the clock at the muxer at the instant the bits comprising
this field left the muxer output.  The time is measured in units of
27MHz from an arbitrary reference point, i.e. there is no connection
with real time.  An audio sample or video frame shall be presented
when the PCR has a value equal to its PTS.  For most frames/samples
the presentation time will be calculated from the most recent PTS
value in the stream and the frame/sample rate.  A local clock is used
to interpolate between PCR values recorded in the stream.

-- 
Måns Rullgård
mru@mru.ath.cx