Recherche avancée

Sur d’autres sites (9041)

• Concatenating two AAC decreasing number of frames packets

  29 décembre 2019, par Ahmed Hawary

  I am trying to concatenate two m4a (aac) files using the FFmpeg with the following command :

  ffmpeg -f concat -i input.txt -codec copy output.m4a

  the first file number of frames using afinfo on macOS :

  File type ID:   m4af
  
  Data format:     1 ch,  44100 Hz, 'aac ' (0x00000000) 0 bits/channel, 0 bytes/packet, 1024 frames/packet, 0 bytes/frame
  
  no channel layout.
  
  estimated duration: 8.473832 sec
  
  audio bytes: 68931
  
  audio packets: 367
  
  bit rate: 64710 bits per second
  
  packet size upper bound: 391
  
  maximum packet size: 391
  
  audio data file offset: 44
  
  not optimized
  
  audio 373696 valid frames + 2112 priming + 0 remainder = 375808
  
  format list:
  
  [ 0] format:      1 ch,  44100 Hz, 'aac ' (0x00000000) 0 bits/channel, 0 bytes/packet, 1024 frames/packet, 0 bytes/frame
  
  Channel layout: Mono

  The second file metadata :

  File type ID:   m4af
  
  Num Tracks:     1
  
  Data format:     1 ch,  44100 Hz, 'aac ' (0x00000000) 0 bits/channel, 0 bytes/packet, 1024 frames/packet, 0 bytes/frame
  
  no channel layout.
  
  estimated duration: 5.594558 sec
  
  audio bytes: 46077
  
  audio packets: 243
  
  bit rate: 65329 bits per second
  
  packet size upper bound: 340
  
  maximum packet size: 340
  
  audio data file offset: 44
  
  not optimized
  
  audio 246720 valid frames + 2112 priming + 0 remainder = 248832
  
  format list:
  
  [ 0] format:      1 ch,  44100 Hz, 'aac ' (0x00000000) 0 bits/channel, 0 bytes/packet, 1024 frames/packet, 0 bytes/frame
  
  Channel layout: Mono

  the resulted audio files metadate :

  File type ID:   m4af
  
  Num Tracks:     1
  
  Data format:     1 ch,  44100 Hz, 'aac ' (0x00000000) 0 bits/channel, 0 bytes/packet, 1024 frames/packet, 0 bytes/frame
  
  no channel layout.
  
  estimated duration: 14.070998 sec
  
  audio bytes: 122792
  
  audio packets: 607
  
  bit rate: 69696 bits per second
  
  packet size upper bound: 293
  
  maximum packet size: 293
  
  audio data file offset: 40
  
  not optimized
  
  audio 620531 valid frames + 1024 priming + 13 remainder = 621568
  
  format list:
  
  [ 0] format:      1 ch,  44100 Hz, 'aac ' (0x00000000) 0 bits/channel, 0 bytes/packet, 1024 frames/packet, 0 bytes/frame
  
  Channel layout: Mono

  The problem is that the total number of frames should be 367+293 = 610 and the resulted number of frames is 607. And the duration is 14.070998 sec instead of 14.06839 sec

  Any ideas if I am doing anything wrong here ? I need to precisely concatenate the two files without any loss or gain in the input frames.

• FFMPEG optimal cmd for large filter_complex

  19 novembre 2019, par Artem Kuznetsov

  I need to mount complex video scenes from several files.
  Many trims, atrims, overlays, amix.
  Maybe it needs to be divided into several calls ?

  And I use something like this :

  ffmpeg -y -i 1.mp4 -i 2.mp4 -i 3.mp4 -i 4.mp4 -i 5.mp4 -i 6.mp4 -i 7.mp4 -i 8 -i 9.mp4 -i 10.mp4 -i 11.mp4 -i 12.mp4 -i 13.mp4 -f lavfi -i anullsrc \
  
  -filter_complex "[0:v]trim=end=0.456, setpts=PTS-STARTPTS[out0]; \
  
  [0:a]atrim=end=0.456, asetpts=PTS-STARTPTS[a_out0]; \
  
  [0:v]trim=0.456:969.456, setpts=PTS-STARTPTS, scale=iw*180/ih:180[out1]; \
  
  [1:v][out1]overlay=main_w-overlay_w-0:main_h-overlay_h:eof_action=pass[out2]; \
  
  [0:a]atrim=0.456:969.456, asetpts=PTS-STARTPTS[a_out1]; \
  
  [0:v]trim=969.456:974.138, setpts=PTS-STARTPTS[out3]; \
  
  [0:a]atrim=969.456:974.138, asetpts=PTS-STARTPTS[a_out2]; \
  
  [0:v]trim=974.138:1382.138, setpts=PTS-STARTPTS, scale=iw*180/ih:180[out4]; \
  
  [2:v][out4]overlay=main_w-overlay_w-0:main_h-overlay_h:eof_action=pass[out5]; \
  
  [0:a]atrim=974.138:1382.138, asetpts=PTS-STARTPTS[a_out3]; \
  
  [2:a][a_out3]amix=inputs=2:duration=first[a_out4]; \
  
  [0:v]trim=1382.138:1408.739, setpts=PTS-STARTPTS[out6]; \
  
  [0:a]atrim=1382.138:1408.739, asetpts=PTS-STARTPTS[a_out5]; \
  
  [0:v]trim=1408.739:2089.758, setpts=PTS-STARTPTS, scale=iw*180/ih:180[out7]; \
  
  [3:v][out7]overlay=main_w-overlay_w-0:main_h-overlay_h:eof_action=pass[out8]; \
  
  [0:a]atrim=1408.739:2089.758, asetpts=PTS-STARTPTS[a_out6]; \
  
  [3:a][a_out6]amix=inputs=2:duration=first[a_out7]; \
  
  [0:v]trim=2089.758:2112.511, setpts=PTS-STARTPTS[out9]; \
  
  [0:a]atrim=2089.758:2112.511, asetpts=PTS-STARTPTS[a_out8]; \
  
  [0:v]trim=2112.511:2358.528, setpts=PTS-STARTPTS, scale=iw*180/ih:180[out10]; \
  
  [4:v][out10]overlay=main_w-overlay_w-0:main_h-overlay_h:eof_action=pass[out11]; \
  
  [0:a]atrim=2112.511:2358.528, asetpts=PTS-STARTPTS[a_out9]; \
  
  ...
  
  [9:v]trim=18.184:36.184, setpts=PTS-STARTPTS, scale=iw*180/ih:180[out26]; \
  
  [10:v][out26]overlay=main_w-overlay_w-0:main_h-overlay_h:eof_action=pass[out27]; \
  
  [9:a]atrim=18.184:36.184, asetpts=PTS-STARTPTS[a_out18]; \
  
  [9:v]trim=36.184:90.237, setpts=PTS-STARTPTS[out28]; \
  
  [9:a]atrim=36.184:90.237, asetpts=PTS-STARTPTS[a_out19]; \
  
  [9:v]trim=90.237:345.237, setpts=PTS-STARTPTS, scale=iw*180/ih:180[out29]; \
  
  [11:v][out29]overlay=main_w-overlay_w-0:main_h-overlay_h:eof_action=pass[out30]; \
  
  [9:a]atrim=90.237:345.237, asetpts=PTS-STARTPTS[a_out20]; \
  
  [9:v]trim=345.237:414.306, setpts=PTS-STARTPTS[out31]; \
  
  [9:a]atrim=345.237:414.306, asetpts=PTS-STARTPTS[a_out21]; \
  
  [9:v]trim=414.306:585.314, setpts=PTS-STARTPTS, scale=iw*180/ih:180[out32]; \
  
  [12:v][out32]overlay=main_w-overlay_w-0:main_h-overlay_h:eof_action=pass[out33]; \
  
  [9:a]atrim=414.306:585.314, asetpts=PTS-STARTPTS[a_out22]; \
  
  [12:a][a_out22]amix=inputs=2:duration=first[a_out23]; \
  
  [9:v]trim=start=585.314, setpts=PTS-STARTPTS[out34]; \
  
  [9:a]atrim=start=585.314, asetpts=PTS-STARTPTS[a_out24]; \
  
  [out0][a_out0][out2][a_out1][out3][a_out2][out5][a_out4][out6][a_out5][out8][a_out7][out9][a_out8][out11][a_out10][out12][a_out11][out14][a_out12][out15][a_out13][out18][a_out14][out19][silence0][out20][a_out15][out23][a_out16][out24][silence1][out25][a_out17][out27][a_out18][out28][a_out19][out30][a_out20][out31][a_out21][out33][a_out23][out34][a_out24]concat=n=23:v=1:a=1[out][a_out]" \
  
  -map [out] -map [a_out] \
  
  -c:v libx264 -crf 18 -pix_fmt yuv420p \
  
  -c:a aac \
  
  output.mp4

  Is this optimal ? Can this be written somehow differently ?
  Now I have this problem :

  [out_0_1 @ 0x7fb2b8ca96c0] 100 buffers queued in out_0_1, something may be wrong. speed=8.51x      
  
  [out_0_1 @ 0x7fb2b8ca96c0] 1000 buffers queued in out_0_1, something may be wrong.

• How to cut video properly with this ffmpeg python script

  11 décembre 2019, par exoboyyeet

  I am trying to write a python script that breaks a video into chunks of less than 64 mb each. This is my for loop that convert each chunk of broken up video :

  for part in range(parts):
  
      print(start, end)
  
      subprocess.run(f"ffmpeg -i {filename} -vcodec copy -acodec copy -ss {start} -t {end} {newfile}\OutputPart{part}.mp4", shell=True)
  
      start = end
  
      end += partlength

  I defined start to be 0 originally, and the end marker to be the length of each part.

  For example, if the video file is 139 mb and 20 mins long, it takes the size divided by 64 [2.171875] turns it into an integer plus 1 in order to add a third part for any amount of video after the last 64 mb marker. Then, it takes that number of parts and divides the length of the video (in our case 20 mins = 1200 seconds) by the number of parts (the variable set previously as parts) which would get us (1200 seconds divided by 3 parts = ) 400 which is the length each part should be (partlength). Now, it runs a loop for the number of parts, to convert a video with the start point (denoted in the ffmpeg command as -ss) originally 0, and the end point (-t), originally the length of 1 part (in our case 400). after the first run through, to make sure the start and endpoints are correct, it prints the start and end points. All the runs say the correct start and endpoints (0-400 ;400-800 ;800-1200). The first and third files convert perfectly, while the second file of the three has from 400-1200 (it includes the third file).

  Is there a reason why it won’t copy the correct segment ?