Hello BC:=Baptiste Coudurier BC> Andrea Barbieri wrote: BC> > Hello BC> > BC> > BC:=Baptiste Coudurier BC> > BC> > BC> Andrea Barbieri wrote: BC> > BC> > Hello Baptiste BC> > BC> > BC> > BC> > BC:=Baptiste Coudurier BC> > BC> > BC> > BC> > BC> Nicolas Jauffret wrote: BC> > BC> > BC> > On Thu, 29 Mar 2007 15:06:27 +0200, Andrea Barbieri BC> > BC> > BC> > <andrea.barbieri at movingimageresearch.com> wrote: BC> > BC> > BC> >> BC> > BC> > BC> >> try the conversion without video essence: BC> > BC> > BC> >> ffmpeg -i foo.avi -acodec mp3 out.mov BC> > BC> > BC> > BC> > BC> > BC> > => Readable with mplayer (no sound without "-demuxer lavf" parameter) BC> > BC> > BC> > => Readable with MacOS QtPlayer 7.1.5 BC> > BC> > BC> > BC> > BC> > BC> > So the problem seems well to come from the "copy" codec. BC> > BC> > BC> BC> > BC> > BC> No, problem might come from mplayer demuxer, if Quicktime plays it, BC> > BC> > BC> file has great chances to be correct. BC> > BC> > BC> > BC> > the point is that in this case the output is playable by both BC> > BC> > applications, when adding the video 'copy' codec this is not true BC> > BC> > anymore... BC> > BC> BC> > BC> No, the point is mplayer not correctly playing, Quicktime/ffmpeg/xine BC> > BC> play it. BC> > BC> > I'm afraid I made the mistake to cut *now* relevant parts of previous BC> > posts which do explain what I am trying to tell you, please have a look at BC> > those. BC> BC> No, read whole explanation, you focus on a weird demuxer behaviour BC> instead of the real issue. Check file, check info written in atoms, BC> compare against specs, finally quote me specs if that is wrong, I'll BC> fix it for sure. I think we are degenerating... BC> BC> > BC> BC> > BC> > BC> BC> > BC> > BC> >> the simpler the problem the better it is to pinpoint a possible bug BC> > BC> > BC> >> BC> > BC> > BC> >> could it be that the avi parser (unlike the wav parser) is not BC> > BC> > BC> >> filling up properly all the required parameters for the mov writer? BC> > BC> > BC> >> BC> > BC> > BC> >> you can use mpeg4ip or Dumpster (Apple quicktime downloadable tool) BC> > BC> > BC> >> to inspect the problematic movie clip BC> > BC> > BC> > BC> > BC> > BC> > $ ffmpeg -i foo.%d.jpg -i bar.wav -vcodec mpeg4 -acodec mp3 from_jpg.mov BC> > BC> > BC> >> /dev/null 2>&1 BC> > BC> > BC> > $ mp4info from_jpg.mov BC> > BC> > BC> > mp4info version 1.5.0.1 BC> > BC> > BC> > Track Type Info BC> > BC> > BC> > 1 video MPEG-4 Simple @ L1, 8.000 secs, 0 kbps, 640x352 @ BC> > BC> > BC> > 25.000000 fps BC> > BC> > BC> > 2 audio .mp3, 8.568 secs, 64 kbps, 48000 Hz BC> > BC> > BC> > BC> > BC> > BC> > $ ffmpeg -i foo.%d.jpg -i bar.wav -vcodec mpeg4 -acodec mp3 foo.avi > BC> > BC> > BC> > /dev/null 2>&1 BC> > BC> > BC> > $ ffmpeg -i foo.avi -vcodec copy -acodec mp3 from_avi.mov > /dev/null 2>&1 BC> > BC> > BC> > $ mp4info from_avi.mov BC> > BC> > BC> > mp4info version 1.5.0.1 BC> > BC> > BC> > Track Type Info BC> > BC> > BC> > 1 video FMP4, 8.000 secs, 378 kbps, 0x0 @ 25.000000 fps BC> > BC> > BC> > 2 audio .mp3, 8.616 secs, 64 kbps, 48000 Hz BC> > BC> > BC> > BC> > BC> > BC> > That is strange that the 2 sound streams have not the same duration. BC> > BC> > BC> BC> > BC> > BC> Well possible, I need to check. BC> > BC> > BC> BC> > BC> > BC> > That is also strange that the video tags are different. BC> > BC> > BC> BC> > BC> > BC> No, default ffmpeg mpeg4 codec tag in avi in FMP4, muxer by default BC> > BC> > BC> uses tag if it is set, patch welcome though. BC> > BC> > BC> > BC> > I think that the mov writer module is not behaving properly when BC> > BC> > creating the video track since the profile and level information is BC> > BC> > missing as well... BC> > BC> BC> > BC> I believe mp4info just does not recognize FMP4 tag and therefore not BC> > BC> even try to gather info. BC> > BC> > I don't think so... (if you check the source code) BC> BC> Well that was just some guess, check mp4info, again check info in atoms BC> compare against specs and I'll fix it. I am not the one with the problem... but I can easily generate a mov file with the same issues and then point at the deficiencies BC> BC> > BC> BC> > BC> > the initial movie file (generated by the jpeg -> mov conversion) BC> > BC> > also appears to be incorrect/incomplete in terms of video track BC> > BC> > information (sample description atoms) the 0 kbps deducted info is BC> > BC> > the giveaway. BC> > BC> BC> > BC> bitrate is just not set in "esds" atom, if you have specs (14496-1 and BC> > BC> 14496-14, maybe 14496-2 for mpeg4) where it says this is mandatory, BC> > BC> I'll fix it. BC> > BC> > you are mentioning MPEG4 standards which define the MP4 format. BC> BC> AFAIK "esds" atom is defined in MPEG4 (14496) standard. BC> BC> > Unfortunately we are talking about the QuickTime file format (the BC> > grandfather of the MP4 and MJPEG200 ones) which behaves differently BC> BC> MP4/MJ2/3GP are derivations of ISO Media file format (14496-12) BC> I am afraid but a bit of hystory lesson is required the QuickTime file format came first, in the late 90's the ISO MPEG-4 group wanted to adopt a file format for MPEG 4... luckily the final choice was to adopt a derivative of Apple QT file format and not Microsoft ASF file format. if you still don't believe me check: http://en.wikipedia.org/wiki/QuickTime BC> > http://developer.apple.com/reference/QuickTime/ BC> BC> Point me where to find "esds" atom description please. http://developer.apple.com/reference/QuickTime/ then follow File Format Specification best regards, andrea -- Andrea Barbieri Moving Image Research http://www.movingimageresearch.com/ The Workshop, Hampton Lane, Bristol, BS6 6LE, UK Tel: +44 117 9732200 FAX: +44 117 9732210 PGP ID=0x034DFD5A KeyFingerprint=C168EA9A7189538D214F1281A7529F32034DFD5A