[MEncoder-users] Mencoder does not cut exactly

belcampo belcampo at zonnet.nl
Thu May 1 18:24:00 CEST 2008


belcampo wrote:
> BUIRA Etienne wrote:
>> On Wed, Apr 30, 2008 at 02:35:43PM +0200, TAJTHY Tammms wrote:
>>> Hi,
>>>
>>> I have a strange problem. I have a mjpeg avi file and I'd like to cut a part
>>> from it and convert to mpeg4. I detected the frames by VirtualDub. I did the
>>> following:
>>>
>>> mencoder ..compression-arguments.. -ss 1.120 -endpos 1:23:00.000 -o cut.avi
>>> patch.avi;
>>>
>>> My problem is that the resulted file does not contain the exact frames.
>>> Depending on the -ss option it can be +/-1 frames different. At the end
>>> position can be +/- 4 frames different. If I add 3-5 ms to the -ss and
>>> -endpos options so the start become puntual, but the problem still appears
>>> at the end.
>> Hi.
>>
>> The -ss and -endpos method is quite inacurrate (even if you got a very
>> good accuracy for using those IMHO...).
>>
>> You'd better use -hr-edl-seek, after selecting good timestamps using
>> -edlout. From the manpage: See http://www.mplayerhq.hu/DOCS/HTML/en/edl.html
>> for details.
>>
>> (Sorry if this message appears twice, I messed up again with my email @)
>> _______________________________________________
>> MEncoder-users mailing list
>> MEncoder-users at mplayerhq.hu
>> https://lists.mplayerhq.hu/mailman/listinfo/mencoder-users
> video and audio are always off by ... msecs, sometimes earlier, 
> sometimes later, depending on source.
> If sound is off by say .4 sec. then 1 video frame will be added. So I 
> think that if you want to be frame-accurate, you'll need to decode video 
> and audio separate. You'll have to start at an I-frame till another 
> I-frame, not including it.
> 
> _______________________________________________
> MEncoder-users mailing list
> MEncoder-users at mplayerhq.hu
> https://lists.mplayerhq.hu/mailman/listinfo/mencoder-users
I wanted distributed encoding, dividing a transcode job to several 
compters, so I also had the frame-accuracy problems and then five-fold. 
Solved it by using ffmpegs ss in combination with the vframe function 
and the audio part in a separate run. And then muxing the video and 
audio parts together. Although I use mp4 format I think you can put your 
parts together again in avi.
I do MP4Box -delay 1=500 -add 1stvideo.h264 -add wholeaudio.aac -cat 
2ndvideo.h264 -cat ...video.h264 frameaccuratewithaudioinsync.mp4




More information about the MEncoder-users mailing list