[x264-devel] Issues with intra refresh?

[Robert Logue]

Yes it is a single threaded encode. I am using libx264.so.120. I was 
initially using gstreamer to transcode videos when I discovered this 
problem, I would hope gstreamer would re initialize the variables. I 
have only altered the x264.c file to reproduce the issue without any 
other component in the way. I am not a x264 programmer so not sure if I 
have re-initialized the variables correctly, it could be I haven't. 
That's why I am here to see if it is a bug or an issue with something I 
am doing.

Robert


On 13/01/12 18:34, Toby wrote:
> Question is similar to the deterministic discussion here:
> http://forums.hexus.net/pc-hardware/118246-x264-video-encoding-benchmark-4.html
>
> For ref, veryfast USED to mean:
> subme 1, ref 1, me dia, partitions i8x8,i4x4, trellis 0, weightp 0,
> no-mixed-refs, no-mbtree
>
> This appears to be a single threaded encode, correct?
>
> What version of libx264? Are you re-initializing every variable that
> remains across your two encodes?
>
> On Thu, Jan 12, 2012 at 11:51 AM, Robert Logue
> <robert.logue at openwave.com>  wrote:
>> I forgot to add I am using libx264.so.120
>>
>>
>> On 12/01/12 16:37, Robert Logue wrote:
>>> Hello,
>>>
>>> When I was running a test harness to go to youtube and download random
>>> videos and transcode I found an issue. The issue is with intra-refresh
>>> turned on libx264 gives back different video for the same input video
>>> and params. This only seems to occur when the 1 process does multiple
>>> transcodes of the same file.
>>>
>>> To test this I altered the x264.c file to transcode the same input file
>>> twice and as output I got two different video files different file size
>>> and different frame data.
>>>
>>> The command I used was
>>>
>>> /spare/x264/install/bin/x264 --preset=veryfast --bitrate=300
>>> --tune=zerolatency --profile=baseline --intra-refresh --vbv-maxrate=300
>>> --vbv-bufsize=12000 --ipratio=0.4 --qpmin=10  -o ./output.mp4 ./input.mp4
>>>
>>> and the video can be found here
>>>
>>> http://199.91.152.89/nhnunbt586kg/5xzsdnshfcgj459/tram.mp4
>>>
>>> Note the video was downloaded by a random test harness and is quiet
>>> offensive. If possible it should not be watched and if watched the sound
>>> should be turned down.
>>>
>>> I was on the web chat earlier and the guys on there indicated that this
>>> could be a bug. Is there any where else to raise this or is here ok?
>>>
>>> Thanks
>>>
>>> Robert
>>> _______________________________________________
>>> x264-devel mailing list
>>> x264-devel at videolan.org
>>> http://mailman.videolan.org/listinfo/x264-devel
>>
>> _______________________________________________
>> x264-devel mailing list
>> x264-devel at videolan.org
>> http://mailman.videolan.org/listinfo/x264-devel
> _______________________________________________
> x264-devel mailing list
> x264-devel at videolan.org
> http://mailman.videolan.org/listinfo/x264-devel