[Ffmpeg-devel] Re: [PATCH] mov fps fix

bond b-o-n-d
Tue Jan 17 17:10:58 CET 2006

> Maybe, checking the last user question about fps in MOV. His file,
> test01.mov had this specifications :
> stts atom:
>        count 10576 duration 24
>        count 1 duration 86
> while timescale is 600.
> Obviously frame rate is 25. But because of that last sample, it cannot
> be computed right. Do you have any idea ?

strictly speaking that file is a variable framerate stream (propably its a
bug in the tool that created the file, but this doesnt change that its vfr)

btw there is another framerate issue with mov/mp4:

the demuxer doesnt seem to handle 64bit values in the mp4
headers correctly, as allowed by the specs (mov?):
the framerate is far off, it should be 29.97 but
15712911.000 fps get used. i guess thats maybe because the duration is not
recognized as being 64bit? (timescale is always 32bit)

attached the structure of the mdhd atom in the case of 64bit times:

   type mdia
    type mdhd
     version = 1 (0x01)
     flags = 0 (0x000000)
     creationTime = 3218293987 (0x00000000bfd344e3)
     modificationTime = 3218295051 (0x00000000bfd3490b)
     timeScale = 15712911 (0x00efc28f)
     duration = 4525654016 (0x000000010dc00000)
     language = 21956 (0x55c4)
     reserved = <2 bytes> 00 00

the version = 1 flag indicates that 64bit is used. 64bit is allowed as can
be seen in the
specs on page 20:

More information about the ffmpeg-devel mailing list