Recherche avancée

Sur d’autres sites (6362)

• Trouble with the ffmpeg -ss flag when capturing one frame from a Macbook iSight webcam

  1er mars 2014, par Andy

  For about five years I've used ffmpeg in a shell script to grab one frame from my linux'd-up Macbook's iSight :

  ffmpeg -f video4linux2 -s 640x480 -r 15 -i /dev/video0 -an -vframes 1 -vcodec mjpeg -y -sameq -ss 1.5 snapshot.jpg

  I just upgraded my Ubuntu distro from Lucid to Natty (ffmpeg ver. 0.6.2-4:0.6.2-1ubuntu1).
  Now that syntax turns the iSight on but hangs indefinitely.

  [output snipped, ending with:]
  
  frame=    0 fps=  0 q=0.0 Lsize=      -0kB time=10000000000.00 bitrate=  -0.0kbits/s    
  
  video:0kB audio:0kB global headers:0kB muxing overhead -inf%

  Without the '-ss' flag it seems to successfully grab the first frame and exit immediately - the only difference in output being :

  frame=    1 fps=  0 q=0.0 Lsize=      -0kB time=0.07 bitrate=  -2.6kbits/s    
  
  video:16kB audio:0kB global headers:0kB muxing overhead -100.132730%

  However, the '-ss 1.5' was necessary to delay the frame capture by 1.5 seconds to allow the cam sufficient time to adjust the exposure.

  The -itsoffset flag seemed promising, but doesn't seem to change ffmpeg's behavior (ie doesn't hang, but no delay).

  Any ideas ?

• Recommendataions for robust and invisible video watermarking software / library [on hold]

  1er mars 2014, par id128

  I am doing a research project in search of a robust and invisible video watermarking software / library (preferably in C++). It needs to meet the following requirements :

  • Robust - that is the watermark needs to survive common transformations such as re-encoding, A/D / D/A conversion, luminance/color change, shift, and crop (both in size and length)'

  • Invisible - to the human eye

  The watermark does not need to carry too much data - maybe a 64-bit UUID - and can be temporal or spatial. But I will need the ability to determine the existence (or not) of the watermark with only a few seconds of the video

  Some research I have done so far :

  • JAWS research paper (http://pdf.aminer.org/000/316/002/the_viva_project_digital_watermarking_for_broadcast_monitoring.pdf) but cannot find any implementations on the web and the authors are not responding to emails. Plus, that research is over 10 years old and I am hoping for something more modern.

  • OpenPuff - does not satisfy the robustness requirement in that the resulting video does not survive transformations.

  • ffmpeg software - I have figured out how to overlay a visible watermark, but that does not satisfy the invisibility requirement.

  Any ideas / suggestions / pointers will be awesome. Thanks.

• avformat/utils : dvd still frames read thru libdvdnav ended up in internal lavf buffer

  2 février 2014, par Voyager1

  avformat/utils : dvd still frames read thru libdvdnav ended up in internal lavf buffer
  This is the solution we’ve been using in XBMC for over 2 years for dvd still frames.
  
  The problem is that the demuxer asks for probing of the codec in the mpeg stream.
  
  This causes lavf to read the whole menu structure into internal buffers.
  
  After which, it won’t read from input stream anymore and no events triggers.
  Signed-off-by : Michael Niedermayer <michaelni@gmx.at>
  

  • [DH] libavformat/utils.c