[Ffmpeg-devel] mpeg2 muxer: lpcm issue

[Måns Rullgård]

Chris said:
> M?ns said:
>
>>>I'll have a look as soon as I  have access to something with a little more
> power than a Nokia 770.

Oops, I forgot all about this one after I got back.  Sorry about that.

> M?ns,
>
> I have been looking at this and come up with the following information which
> may help you with any investigation:
>
> Firstly I ran the output mpeg through Phillips DVD-Video verifier which
> spewed out many DVD/Mpeg issues but a few were interesting to this
> problem...
>
> First: [DVD] ERROR 3235  (ref. DVD-3 Fig. 5.2.3/4/5-1)
>
> (see:
> http://www.licensing.philips.com/includes/download.php?id=6106&filename=2498
> .pdf)
>
> Search for "ERROR 3235"
>
> This error would indicate that the max payload for LPCM PES packets is 2017
> where as the output mpeg has 2019...
>
> Also: [DVD] ERROR 3870  (ref. DVD-3 Table 5.2.4-1, note 3) :
>     The number_of_frame_headers (7) must describe the number of audio frames
> (6)
> 	whose first byte is in this A_PKT
>
> And: [DVD] ERROR 3871:
> 	The first_access_unit_pointer (value 4) must describe the address
> 	of the first byte of the first AU in this A_PCK
>
>
> The first "error" can be sorted by reducing the payload_size and adjusting
> the stuffing. The last two issues pointed me to investigate the audio
> substream headers in dvd's (please see attached - if it comes through!).
>
> The issue here is that the first 3 bytes of the header (frame count, and
> offset to first access unit) are always the same in mpeg.c yet I assume they
> need to be dynamic. Also FrmNum in the fourth byte should be dynamic too.

I figured it might be something with the packet headers.

> It is unclear to me whether these are purely DVD Vob issues or mpeg 2
> program stream issues too.

ISO MPEG2 doesn't specify storage of lpcm audio.  It is a dvd extension.

-- 
M?ns Rullg?rd
mru at inprovide.com