Recherche avancée

Sur d’autres sites (10327)

• New Sanyo cameras have editing in mind

  13 octobre 2009

  Sanyo has announced some ’A’ revisions to their existing FH1 and HD2000 cameras, which add a new "iFrame" mode. It appears this is an i-frame only h264 mode, at a reduced 960x540 resolution. It’s a very interesting idea - if other manufacturers adopted it as an optional setting, and if NLE manufacturers supported it, it could turn H264 into an edit-friendly format. Right now, editing H264 is hamstrung by the extremely long GOPs and complex interframe relationships. Going to i-frame only makes it essentially a more advanced version of a codec like DV or DVCProHD.

  Interestingly, the bottom of the press release mentions that

  "The iFrame logo and the iFrame symbol are trademarks of Apple Inc."

  That’s news to me. One wonders if Sanyo jumped the gun on a release, or if this is just a format that Apple uses internally in tools like iMovie, which Sanyo has co-opted. I’ll certainly be keeping my eyes open for an Apple announcement about "iFrame."

• FFMPEG Select audio stream by Codec or Bitrate

  25 juillet 2018, par CRS

  I’m trying to convert some TV recorded videos with ffmpeg.
  Target format should be : MP4/H264/AAC
  Most movies are in TS/MPEG2/MPEG2 audio

  Some movies also have an AC3 stereo or AC3 surround track.

  I want the ffmpeg to always prefer the AC3 track as a source if it exists because this usually has a better quality. If the Video is multichannel, that works anyway since ffmpeg automatically selects the track with the most channels. But if it is only 2 channels, often the MPEG2 Audio Track is used.

  Of course, I could achieve that for each movie individually with the map option, but I want to convert a few hundred movies by batch and the stream IDs are mixed thoroughly so I want ffmpeg to automatically select either the AC3 Track or the Track with the highest bitrate for conversion.

  Another Solution would be : Always convert the MPEG Stream to AAC and copy the AC3 one to the output file if it exists. I just don’t want to lose the AC3 Stream.

• FFMPEG vs Gstreamer, performance/filesize ?

  10 février 2021, par kepitto

  I'm looking for a tool to record a udp stream and store it as efficiently as possible. So far I've tried gstreamer and ffmpeg. They key difference I noticed so far is that ffmpeg uses way less storage space while using the same codec.

  


  I capture a mjpeg stream and convert it to x264. Shouldn't x264 be the same compression regarding size accross platforms ?

  


  Following commands were run :

  


  


  ffmpeg -i udp ://192.168.2.100:52000 -c:v libx264 -c:a aac out.mp4

  


  


  filesize : 2,4MB/s
  
avg cpu usage : 68%

  


  


  gst-launch-1.0 udpsrc port=52000 do-timestamp=true
caps=image/jpeg,framerate=30/1 blocksize=4096 ! queue ! jpegparse !
queue ! jpegdec ! queue ! videoconvert ! x264enc ! queue ! matroskamux
 ! filesink location=x.mkv

  


  


  filesize : 12,5MB/s
  
avg cpu usage : 95%

  


  There's no visual difference which get's noticed with bare eyes between the two videos.

  


  Why is there such a huge difference ? Anything wrong with my pipelines or is ffmpeg just a better tool ?
  
Extra question : When doing these types of encoding, if i have an h264 hardware encoder on my cpu, is it used automatically or do I have to do something in order to enable that ?