[x264-devel] Re: non monotone timestamps...

[Baptiste COUDURIER]

Hi

bond wrote:
>> On Sun, Aug 20, 2006 at 02:04:22PM -0700, Loren Merritt wrote:
>>> On Sun, 20 Aug 2006, Daniel J Blueman wrote:
>>>
>>>> When performing 2-pass encoding [1], I see many "non monotone
>>>> timestamps" errors [2], so frames are dropped, resulting in a shorter
>>>> video stream than the audio stream.
>>>>
>>>> I tracked this down to when the +bpyramid option is passed to x264; no
>>>> problem occurs with '-bpyramid', ie when using B-frames as references
>>>> is disabled. I'm converting a MJPEG 640x480 30 fps video from my
>>>> digital camera with ffmpeg etc trees from late Jul 06 subversion.
>>>>
>>>> Doing a quick search, it seems little is known about this issue. Any
>>>> ideas? I'm not on the mailing list, so please CC.
>>> libavformat doesn't support bpyramid.
>> do you know why? no i am not aware of an issue with bpyramid and lavf
>> just mencoder + bframes
> 
> the ctts atom isnt created correctly when using b-pyramid with x264
> outputting to .mp4, therefore ffmpeg is not able to create spec compliant
> mp4 files when using b-pyramid (normal b-frames are fine)
> 
> x264's own tool handles this fine
> 

'ctts' atom is actually created correctly. If a problem resides, it is
most likely cause pts and/or dts of the packets are incorrect.

Problem with B pyramid is just that lavf will not let packets pass since
they have non monotone timestamps as mentioned in the first post of the
thread.

-- 
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
SMARTJOG S.A.                                    http://www.smartjog.com
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
Phone: +33 1 49966312

-- 
This is the x264-devel mailing-list
To unsubscribe, go to: http://developers.videolan.org/lists.html