Recherche avancée

Sur d’autres sites (7207)

• ffmpeg get last x seconds with high accuracy

  12 mars 2018, par rbarab

  I would like to batch process mp4 videos, getting the last x seconds of each and saving them to individual files.
  I need to do this with a very high accuracy, preferably to 0.001 seconds or better.
  Found a related question (FFMPEG : get last 10 seconds) suggesting -sseof, which works great, but as the answer said it’s not completely accurate with stream copy.

  I am trying to match video lengths to the length of a reference video.

  Would I need to re-encode ? Can sseof handle this accurate enough if I specify duration as 00:00:00.000000 (which I get from reference video ffprobe) ?

  Please see related ffprobe -i below, all videos to be processed have this same encoding.

     Metadata:
  
      major_brand     : isom
  
      minor_version   : 512
  
      compatible_brands: isomiso2avc1mp41
  
      encoder         : Lavf57.83.100
  
    Duration: 00:00:58.67, start: 0.000000, bitrate: 639 kb/s
  
      Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p, 640x360, 499 kb/s, 29.97 fps, 29.97 tbr, 30k tbn, 59.94 tbc (default)
  
      Metadata:
  
        handler_name    : VideoHandler
  
      Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 131 kb/s (default)
  
      Metadata:
  
        handler_name    : SoundHandler
  
  duration=58.673000

  Is there a better way to achieve frame-level accuracy ? As end goal I would need to overlay these videos with 25fps ’frame-level accuracy’.

  Thanks a lot !

• Revision 805078a1bf : build : convert rtcd.sh to perl significantly speeds up file generation. the go

  24 février 2014, par James Zern

  Changed Paths :
   Add /build/make/rtcd.pl

  
   Delete /build/make/rtcd.sh

  
   Modify /libs.mk

  
   Add /vp8/common/rtcd_defs.pl

  
   Delete /vp8/common/rtcd_defs.sh

  
   Modify /vp8/vp8_common.mk

  
   Add /vp9/common/vp9_rtcd_defs.pl

  
   Delete /vp9/common/vp9_rtcd_defs.sh

  
   Modify /vp9/vp9_common.mk

  
   Modify /vpx_scale/vpx_scale.mk

  
   Add /vpx_scale/vpx_scale_rtcd.pl

  
   Delete /vpx_scale/vpx_scale_rtcd.sh

  
  
  build : convert rtcd.sh to perl

  significantly speeds up file generation.

  the goal of this change is to convert rtcd.sh to perl as directly as
  possible to allow for simple comparison. future changes can make it more
  perl-like.

  — -
  Linux
  [CREATE] vpx_scale_rtcd.h
  real 0m0.485s -> 0m0.022s
  [CREATE] vp8_rtcd.h
  real 0m4.619s -> 0m0.060s
  [CREATE] vp9_rtcd.h
  real 0m10.102s -> 0m0.087s

  Windows
  [CREATE] vpx_scale_rtcd.h
  real 0m8.360s -> 0m0.080s
  [CREATE] vp8_rtcd.h
  real 1m8.083s -> 0m0.160s
  [CREATE] vp9_rtcd.h
  real 2m6.489s -> 0m0.233s

  Change-Id : Idfb71188206c91237d6a3c3a81dfe00d103f11ee

• Data lost in ffmpeg in overlay / steganography using ffmpeg

  3 août 2020, par g Kishore

  Goal : Embedding some textual information into a video(steganography) using LSB encoding. And retrieving the embedded text.

  


  Procedure followed :

  


  

  1. Extracted one frame from video :
ffmpeg -i -vf "select=eq(n,34)" -vframes 1 out.bmp

  


  2. LSB encoding of text into out.bmp using C-code :
NOTE : Decoding of embedded text from out.bmp is working fine[c-code]

  


  3. Insert out.bmp back into original video :
ffmpeg -i -i out.bmp -filter_complex "[0:v][1:v]overlay=enable='between(n,0,0)'[out]" -acodec copy -map
"[out]" -map 0:1 output.mp4

  


  4. Extracted frames from encoded video "output.mp4" :
ffmpeg -i output.mp4 frames%d.bmp
NOTE : Replaced image is present in extracted frames

  


  5. Decoding the text from frames0.bmp, frames1.bmp, frames2.bmp using C-Code
NOTE : not able to decode the text. I tried comparing image data of out.bmp & frames0.bmp, frames1.bmp, frames2.bmp. Lots of changes are there.

  


  


  PS :
. C-code used to encode & decode text into the image is working well. No ambiguity in it.
. Encoded video is playable properly. Only issue is loss of embedded text while overlaying the encoded image back into the video

  


  Any help is appreciated.