[NUT-devel] Re: [Ffmpeg-devel] Re: integrating AVS decoding into MPlayer

[Måns Rullgård]

Baptiste Coudurier <baptiste.coudurier at smartjog.com> writes:

> Hi
>
> Michael Niedermayer wrote:
>> [...]
>> 
>>>>> I persist to say that a container SHALL standardize a codec
>>>>> wraping (even just saying "default wraping rule applies" and
>>>>> its fourcc is ...) if it needs to support it. It should just be
>>>>> a matter of one line per codec, as NUT is very generic ;)
>>>> The one line would be the same for every codec so it's pointless.
>>>> 
>>> IMHO not for H264,
>> 
>> what exactly is unclear in the h.264 spec?
>
> ISO/IEC specify bytestream NAL formating used in AVI.
> 1) Wraping in AVI IS NOT standardized.
> 2) Wraping in MOV IS standardized in ISO 14496-15.
>
> Problem will occur with "stream copy" and reformating. Im just saying
> that one way MUST be chosen, and clearly mentioned somewhere.

There is also the possibility of allowing both.  It is easy enough for
the decoder to detect which was used.

>>>> What about timecode ? Subtitles ?
>>> There is no specifications in NUT specs about wraping those kind of
>>>  data. A stream type identifier is not sufficient IMHO.
>>> 
>>> I just read the specs again and I also do not see and decent
>>> standard to pack extradata for specific codec. Just put the MOV
>>> atoms in NUT ?
>> 
>> which codec spec is unclear about how to store it? ogg based codecs
>> yes, that needs a line to clarify
>
> AAC, QDM2, ALAC, MP3, basically everything that uses an extradata IMHO.
> Atm libavcodec assumes extradata is stored as MOV atoms. That is ugly
> IMHO and making NUT to forbid those format is needed.
>
> I think that we should mention wraping in order to be more specific and
> avoid "blind" stream copy for those codecs.
>
> For example, no extradata needed for AAC since it is stored as ADTS.
> or AAC must be stored as frame and ADTS header as extradata.

Don't forget about ADIF.

-- 
Måns Rullgård
mru at inprovide.com