[Mp4-tech] about MPEG2 stream T-SDT

[Samuel Rivas]

Song wrote:
>    Dear experts,
> 
>    I am trying to find the relationship between PCR, DTS and vbv_delay.
>    I can find the following description at the section 2.4.2.3 of ISO13818-1.
>    This is about T-SDT.
> 
>    When the vbv_delay method is used, the final byte of the video picture
>    start code for picture j is transferred from MBn to
>    the EBn at the time tdn(j) šC vbv_delay(j), where tdn(j) is the decoding
>    time of picture j, as defined above, and
>    vbv_delay(j) is the delay time, in seconds, indicated by the vbv_delay
>    field of picture j.
> 
>    Can I reach a conclusion that the PCR at the start code of picture
>    data equate to DTS - vbv_delay?
>    I checked a lot of MPEG2 streams to verify this, but vbv_delay always are
>    too large. Can anyone explain me about the relationship?

  The buffer model defined in the standard uses three buffers for
MPEG-2 elementary streams:
  --> TB --> MB --> EB --> Decoder
  When vbv_delay method is used, vbv_delay is the time that the data
need to traverse EB, but you stil have to count the time in TB and MB.
  Thus, the relationship between PCR, DTS, and vbv_delay is:
  DTS(j) = vbv_delay(j) + MB_delay(j) + TB_delay(j) + t(j)
  Where t(j) is the arrival time of the picture to the TB. It is derived from
the PCR assuming constant bit rate between each pair of PCR values.
  Anyway, I am afraid that this does not explain why your vbv_delays
are too large (unless you confused the last PCR value with t(j)), they should
have been too short ... Have you checked that the vdb_delay method is set in
your transport streams? Otherwise, the leak method is used.
  Regards.
-- 
	Samuel