[FFmpeg-devel] [RFC][PATCH] ticks_per_frame / timebase

[Baptiste Coudurier]

Baptiste Coudurier wrote:
> Michael Niedermayer wrote:
>> On Thu, Feb 26, 2009 at 02:46:36PM -0800, Baptiste Coudurier wrote:
>>> Michael Niedermayer wrote:
>>>> On Thu, Feb 26, 2009 at 02:18:41PM -0800, Baptiste Coudurier wrote:
>>>>> Hi guys,
>>>>>
>>>>> Michael Niedermayer wrote:
>>>>>> On Thu, Feb 26, 2009 at 10:05:28PM +0100, Ivan Schreter wrote:
>>>>>>> Michael Niedermayer wrote:
>>>>>>>> [...]
>>>>>>>> also i think the reset of the sei vars should be in
>>>>>>>> decode_nal_units()
>>>>>>>> ...
>>>>>>>>     if(!(s->flags2 & CODEC_FLAG2_CHUNKS)){
>>>>>>>>         h->current_slice = 0;
>>>>>>>>         if (!s->first_field)
>>>>>>>>             s->current_picture_ptr= NULL;
>>>>>>>>         <-----------------------here
>>>>>>>>     }
>>>>>>>>
>>>>>>>> the reason is that just reseting at the end of a pic is "risky" if that 
>>>>>>>> end
>>>>>>>> isnt reached
>>>>>>>>
>>>>>>>>   
>>>>>>> Like in the attached patch?
>>>>>> yes, patch ok
>>>>>>
>>>>>> btw, if you want a svn write account, (and agree to the devel/svn policy
>>>>>> and split patches a little more, this one should be
>>>>>> factorizing the code and then seperately adding the new call)
>>>>>> then send diego b. some username & password gpg encrypted
>>>>>>
>>>>> Applied both patches:
>>>>>
>>>>> ffmpeg -i h264.mp4 -vcodec copy test.mov
>>>>>
>>>>> Seems stream 1 codec frame rate differs from container frame rate: 12.50
>>>>> (25/2) -> 25.00 (25/1)
>>>>> Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'H264.mp4':
>>>>>   Duration: 00:00:02.18, start: 0.000000, bitrate: 2245 kb/s
>>>>>     Stream #0.0(eng): Audio: aac, 44100 Hz, stereo, s16
>>>>>     Stream #0.1(eng): Video: h264, yuv420p, 720x576, 25 tbr, 25 tbn, 25 tbc
>>>>> File 'test.mov' already exists. Overwrite ? [y/N] y
>>>>> Output #0, mov, to 'test.mov':
>>>>>     Stream #0.0(eng): Video: libx264, yuv420p, 720x576, q=2-31, 90k tbn,
>>>>> 25 tbc
>>>>>     Stream #0.1(eng): Audio: libfaac, 44100 Hz, stereo, s16, 64 kb/s
>>>>>
>>>>> Is this supposed to report 12.50 ?
>>>> no, ive droped the hunk that messed with the printing in ffmpeg.c locally
>>>> is the nonsense it prints the only problem? or are there other problems?
>>>>
>>> It seems ok, however it does not do the same for MPEG-2 when I stream copy:
>>>
>>> Seems stream 0 codec frame rate differs from container frame rate: 25.00
>>> (50/2) -> 25.00 (25/1)
>>> Input #0, mpeg, from 'PubMatrix.mpg':
>>>   Duration: 00:00:24.26, start: 0.220000, bitrate: 11888 kb/s
>>>     Stream #0.0[0x1e0]: Video: mpeg2video, yuv422p, 720x608 [PAR 1:1 DAR
>>> 45:38], 11386 kb/s, 25 tbr, 90k tbn, 50 tbc
>>>     Stream #0.1[0x1c0]: Audio: mp2, 48000 Hz, stereo, s16, 384 kb/s
>>> File 'test.mov' already exists. Overwrite ? [y/N] y
>>> Output #0, mov, to 'test.mov':
>>>     Stream #0.0: Video: mpeg2video, yuv422p, 720x608 [PAR 1:1 DAR
>>> 45:38], q=2-31, 11386 kb/s, 90k tbn, 50 tbc
>>>     Stream #0.1: Audio: libfaac, 48000 Hz, stereo, s16, 64 kb/s
>>> Stream mapping:
>>>   Stream #0.0 -> #0.0
>>>   Stream #0.1 -> #0.1
>>>
>>> In this case pkt->duration is 2 and timebase.den is 50, with H.264
>>> pkt->duration is 1 and timebase.den is 25, is this normal ? Can't we get
>>> 1 and 1/25 with MPEG-2 too ?
>> yes, the following hunk does that:
>> @@ -1750,9 +1751,10 @@
>>              codec->bit_rate = icodec->bit_rate;
>>              codec->extradata= icodec->extradata;
>>              codec->extradata_size= icodec->extradata_size;
>> -            if(av_q2d(icodec->time_base) > av_q2d(ist->st->time_base) && av_q2d(ist->st->time_base) < 1.0/1000)
>> +            if(av_q2d(icodec->time_base)*icodec->ticks_per_frame > av_q2d(ist->st->time_base) && av_q2d(ist->st->time_base) < 1.0/1000){
>>                  codec->time_base = icodec->time_base;
>> -            else
>> +                codec->time_base.num *= icodec->ticks_per_frame;
>> +            }else
>>                  codec->time_base = ist->st->time_base;
>>              switch(codec->codec_type) {
>>              case CODEC_TYPE_AUDIO:
>>
>>
>> more issues left before i can commit it?
> 
> Nice.
> 
> Please gimme 5 mins, I'm testing TS with H.264 and MPEG-2.
> 

I don't see any regression for now.

-- 
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
checking for life_signs in -lkenny... no
FFmpeg maintainer                                  http://www.ffmpeg.org