[FFmpeg-devel] [PATCH] decode at least 4 H.264 frames in av_find_stream_info

[Michael Niedermayer]

On Sun, Jun 27, 2010 at 12:39:13AM -0700, Baptiste Coudurier wrote:
> On 6/26/10 7:43 PM, Michael Niedermayer wrote:
>> On Sun, Jun 27, 2010 at 04:31:41AM +0200, Michael Niedermayer wrote:
>>> On Thu, Jun 24, 2010 at 05:44:20PM -0700, Baptiste Coudurier wrote:
>>>> Hi guys,
>>>>
>>>> $subject.
>>>>
>>>> This will permit the decoder to compute has_b_frames correctly when
>>>> b-pyramid is present. 4 is typically the minimum needed, though maybe we
>>>> should decode more based on has_b_frames. Any opinion ?
>>>
>>> Finally found an old patch by myself doing this
>>> I think its a bit more readable
>>> if it works feel free to apply mine (assuming that thing still applies 
>>> and
>>> works)
>>
>> ehm, the patch:
>>
>> Index: libavformat/utils.c
>> ===================================================================
>> --- libavformat/utils.c	(revision 18646)
>> +++ libavformat/utils.c	(working copy)
>> @@ -1820,6 +1820,12 @@
>>   #endif
>>   }
>>
>> +/*we need to find has_b_frames approximatly, the parser is not yet able 
>> to do it*/
>> +static int has_decode_delay_been_guessed(AVCodecContext *enc)
>> +{
>> +    return enc->codec_id != CODEC_ID_H264 || enc->frame_number>= 4 + 
>> enc->has_b_frames;
>> +}
>
> I'm not sure about enc->frame_number, now that st->codec_info_nb_frames is 
> used in av_find_stream_info, it seems the best field to use.

hmm, i dont think theres a difference currently between using these 2
though they are semantically a bit different. So use what you prefer


>
> Your patch is not enough I think, you need to always call try_decode_frame 
> even if has_codec_parameters is true.

calling it for the first 4 frames with h264 is ok
calling it unconditionally always would slow av_find_stream_info() down quite
a bit i think

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Awnsering whenever a program halts or runs forever is
On a turing machine, in general impossible (turings halting problem).
On any real computer, always possible as a real computer has a finite number
of states N, and will either halt in less than N cycles or never halt.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20100627/71e417e5/attachment.pgp>