[FFmpeg-devel] [PATCH] get the right value for the dac3 atom

[Baptiste Coudurier]

Damiano Galassi wrote:
> Il giorno 04/set/08, alle ore 20:18, Baptiste Coudurier ha scritto:
> 
>> Damiano Galassi wrote:
>>> Hi,
>>> mov_write_ac3_tag starts reading the values 2 bytes too soon.
>>> The attached patch fixes it.
>>>
>> Thanks, applied.
>>
>> Much interest into ac-3 in mp4/mov it seems.
>> Btw testing very welcome, I only could test with quicktime player,
>> wether it recognized the track, the weird thing is that it seems to be
>> recognized without the 'dac3' atom.
>>
> 
> The are different ways of muxing ac-3 in mov, all valid. The dac3 atom  
> inside mov has been used only by apple in QuickTime 7.5 when they  
> introduced an ac3 importer.

It's standardized by ETSI now, so there is only one way to mux ac-3 in
mp4. .mov is another story of course, but the official way will always
be apple's one. I could not generate .mov containing ac-3 yet, samples
welcome.

> I think it's more important to test ac-3 in mp4. QuickTime can open it  
> only if the extension is m4v or m4a, and in this case it reads the  
> channels count and the sample rate from the dac3 atom.

.m4v I could generate seems to use specs from ETSI.

> Btw, on OS X and with an ac-3 component installed QuickTime is able to  
> play the files generated by ffmpeg without problems.

Well, it must works "out of the box" without any other component I
think, at least it did on my leopard with compressor 3, quicktime
recognizes the track.

-- 
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
Smartjog USA Inc.                                http://www.smartjog.com
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA