[FFmpeg-devel] [PATCH] simplify ipmovie.c pts calculation

[Mike Melanson]

Michael Niedermayer wrote:
> On Fri, Feb 27, 2009 at 02:49:50PM +0100, Reimar D?ffinger wrote:
>> On Fri, Feb 27, 2009 at 02:36:41PM +0100, Michael Niedermayer wrote:
>>> On Fri, Feb 27, 2009 at 02:27:06PM +0100, Reimar D?ffinger wrote:
>>>> On Fri, Feb 27, 2009 at 02:18:12PM +0100, Reimar D?ffinger wrote:
>>>>> On Fri, Feb 27, 2009 at 01:45:39PM +0100, Michael Niedermayer wrote:
>>>>>> On Fri, Feb 27, 2009 at 08:59:14AM +0100, Reimar D?ffinger wrote:
>>>>>>> On Fri, Feb 27, 2009 at 12:28:03AM +0100, Michael Niedermayer wrote:
>>>>> [concerning code to guess r_frame_rate in libavformat/utils.c]
>>>>>
>>>>>>>> also if you ignore the first 1-2 durations then this might fix the fps of
>>>>>>>> h264_acc_in_flv.flv i think
>>>>>>> I set the code to ignore the first 4, for that file it gives a frame
>>>>>>> rate of 22.2222... which seems not really right either, though I don't
>>>>>>> actually know what the right value is.
>>>>>>> It does not make a difference though, since the get_std_framerate-based
>>>>>>> code after it overwrites it...
>>>>>> fix that prevents it from increasing the frame rate is welcome ...
>>>>> Well, I thought the code was also there to prefer standard rates, not
>>>>> just reduce the frame rate.
>>>>> I have not extensively tested, but does attached patch fit the intended
>>>>> purpose of the code?
>>>> Sorry, here is a fixed version. The "if" looks a bit ugly though.
>>> ok if tested a little
>> Hm, this file:
>> http://download.theforce.net/theater/shortfilms/sinsofthejedi/SinsOfTheJedi_Part1.mov
>> is detected as
>> Stream #0.1(eng): Video: svq3, yuvj420p, 534x400, 14.98 tbr, 29.97 tbn, 29.97 tbc
>> But that seems to be related to your ticks_per_frame changes?
>> tb_unreliable at least is not set for this, so it can't really be any of
>> the code I was working on...
> 
> should be fixed
> 
> [...]
> 
> PS: a SVQ3 regerssion test would be very welcome, i broke SVQ3 so often
>     already with h264 changes ...

I am downloading the above sample. What was the nature of the problem? 
Will it manifest (in timestamps) using "-f framecrc -" encoding target? 
If so, should I take the first 2 MB of the file and add it to the FATE 
suite?

-- 
     -Mike Melanson