Recherche avancée

Sur d’autres sites (4377)

• ffmpeg messing up audio when copying PCM streams

  2 juin 2022, par lioumens

  I'm using ffmpeg to compress videos, but when trying to copy the audio stream from one video file using the same codec and container, the audio is completely corrupted and unusable.

  


  The video I'm trying to compress has stream data :

  


  $ ffprobe -hide_banner -i testing.mov
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'testing.mov':
  Metadata:
    major_brand     : qt
    minor_version   : 0
    compatible_brands: qt
    creation_time   : 2019-04-25T05:37:35.000000Z
    com.apple.quicktime.make: Apple
    com.apple.quicktime.model: MacBookPro15,2
    com.apple.quicktime.software: Mac OS X 10.14.2 (18C54)
    com.apple.quicktime.creationdate: 2019-04-25T00:28:40-0500
  Duration: 00:06:35.75, start: 0.000000, bitrate: 66117 kb/s
  Stream #0:0[0x1](und): Video: prores (Standard) (apcn / 0x6E637061), yuv422p10le(tv, smpte170m/bt709/bt709, progressive), 1280x720, 65923 kb/s, 29.01 fps, 29.01 tbr, 30k tbn (default)
    Metadata:
      creation_time   : 2019-04-25T05:37:35.000000Z
      handler_name    : Core Media Video
      vendor_id       : appl
      encoder         : Apple ProRes 422
  Stream #0:1[0x2](und): Audio: pcm_s16be (lpcm / 0x6D63706C), 8000 Hz, 1 channels, s16, 128 kb/s (default)
    Metadata:
      creation_time   : 2019-04-25T05:37:35.000000Z
      handler_name    : Core Media Audio
      vendor_id       : [0][0][0][0]


  


  Just working with the audio, I tried to copy the audio into a separate file with ffmpeg -i testing.mov -vn -channel_layout mono -c:a copy tmp.mov, which results in a video file with stream data :

  


  $ ffprobe -hide_banner -i tmp.mov
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'tmp.mov':
  Metadata:
    major_brand     : qt
    minor_version   : 512
    compatible_brands: qt
    encoder         : Lavf59.16.100
  Duration: 00:06:35.26, start: 0.018000, bitrate: 128 kb/s
  Stream #0:0[0x1]: Audio: pcm_s16be (twos / 0x736F7774), 8000 Hz, mono, s16, 128 kb/s (default)
    Metadata:
      handler_name    : Core Media Audio
      vendor_id       : [0][0][0][0]


  


  It seems the only difference when copying the audio is the original says pcm_s16be (lpcm / 0x6D63706C) while the resulting audio has pcm_s16be (twos / 0x736F7774). My understanding is that twos is Apple Quicktime's FOURCC identifier for signed 16 bit big-endian., which is correct and shouldn't affect the audio.

  


  It seems another forum user had a similar issue, but they were using a hex editor to directly manipulate bits, which I'm not comfortable doing. Is there a fix from within ffmpeg ?

  


  EDIT :
I've uploaded a sample of the clip that has this problem.

  


• avcodec/pthread_slice : Don't use static variable, fix race

  28 novembre 2020, par Andreas Rheinhardt

  avcodec/pthread_slice : Don't use static variable, fix race
  ff_slice_thread_init() uses a static variable to hold a function
  
  pointer, although the value of said pointer needn't be saved between
  
  different runs of this function at all.
  The reason for this being so is probably that said pointer points to
  
  a static function (if used) ; but storage class specifiers like "static"
  
  are not part of the type of an object and so including it in the pointer
  
  declaration is wrong (anyway, "static" means different things in both
  
  contexts : for the function declaration it affects linkage, for the
  
  variable storage duration).
  Using a static variable here can lead to races, e.g. when initializing
  
  VP9 (for which said function pointer was added) and H.264 with slice
  
  threading. The latter has the FF_CODEC_CAP_INIT_THREADSAFE flag set and
  
  is therefore unaffected by the lock guarding initializations of
  
  decoders.
  Reviewed-by : Anton Khirnov <anton@khirnov.net>
  
  Signed-off-by : Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
  

  • [DH] libavcodec/pthread_slice.c

• ffMPEG write two outputs with one encoder and several filters (segmentation, timecode)

  4 mai 2023, par Francois

  need some help after trying days on this. I'm able to process one input to two outputs with two encodings :

  


  ffmpeg -f dshow -i video="Integrated Camera" -c:v mpeg2video -b:v 50000k -vf "scale=1920:1080" -an -flags +global_header -f segment -segment_time 60 -segment_format mxf -strftime 1 -reset_timestamps 1 -minrate 50000k -maxrate 50000k -bufsize 50000k "C:\ffmpeg\Master\Master_%Y-%m-%d_%H-%M-%S.mxf" -b:v 50000k -vf "scale=1920:1080" -an -flags +global_header -f segment -segment_time 60 -segment_wrap 3 -segment_format mxf -reset_timestamps 1 -minrate 50000k -maxrate 50000k -bufsize 50000k "C:\ffmpeg\Backup\backup_%02d.mxf"


  


  But when i try to do something to get rid of the second encoding and working with "tee" filter, i run into the problem that the other filters will not work anymore. If i run this without the extra filters :

  


  ffmpeg -f dshow -i video="Integrated Camera" -c:v mpeg2video -an -f tee -map 0:v -an "Main.mxf|[f=mxf]Backup.mxf"


  


  This works but trying to add this filters after the -f tee filter will lead into an error. As well to define this before the tee filter will not work due to different filter settings in segmentation.

  


  I think the way how it could work is to split the output and use afterwards the tee filter. But i wasn't able to do. Maybe someone could help me here.

  


  Best

  


  Francois