[Ffmpeg-devel] FFmpeg Roadmap

[Colin Ward]

Tim Allen wrote:
>>   So can someone say once and for all, why is it so difficult to get 
>> correct pts values from .mov files when they work for all other formats?
> 
> I'm not that familiar with the ffmpeg mov demuxer, but I can make a 
> general comment, which is that the frame rate and pts in a mov file tend 
> to be defined in at least two, sometimes three, different places (the 
> wrapper header, the track headers, and sometimes embedded in the tracks 
> themselves), and in practice they don't always agree. I've seen files 
> which have one or the other correct, and the other incorrect, with no 
> real consistency about which is definitive. I've also seen cases where 
> both are approximately correct, but different from each other, and 
> neither seems to be exactly correct. So I'm not in the least surprised 
> that the ffmpeg demuxer has a hard job getting it right.

   I am not surprised about that, as it is an old file format that has 
been refined over the years.  Having assisted in the development of a 
couple of file formats myself I know what it is like to add changes 
while retaining downward compatibility - what a pain!

   But my point is that the algorithm that client code uses to convert 
the information returned by decoding a packet, into a timestamp, works 
for all file formats except .mov and that needs to be fixed, regardless 
of the difficulties that the demuxer/parser has in obtaining that 
information.  Currently I can't get the correct timestamps for *any* 
Quicktime movie!

-- 
/-------------------------------------------------------------------\
[Hitman/Code HQ - 6502/z80/68000/604e/80x86/ARM coder - Amiga rulez!]
[VZ-200/VIC-20/MZ-700/c16/c64*10/c128*8/Plus-4/CPC464/CD32/500*2    ]
[600/1000/1200*2/A4000/SNES/N64/Dreamcast/Athlon 1100/AmigaOne      ]
[Assembly Language: The most fun you can have with your clothes on! ]
\-------------------------------------------------------------------/