concatenating MP4 files reads all of them twice

[AyrA]

Summary of the bug

When concatenating multiple mp4 files using -f concat each input file is read twice but only on the second pass the file is actually processed. This problem does not seem to be present for other formats like AVI or MKV.

The generated output file looks as expected and works.

How to reproduce

Command: ffmpeg -f concat -safe 0 -i list.txt -c copy -strict experimental output.mp4

list.txt: File with at least one mp4 file in it

Other Info

• My device runs Microsoft Windows 7 x64 (6.1.7601)
• The first seemingly useless pass is slower than the second actual pass but CPU utilization stays almost at 0 during the entire time of the first pass (Unbuffered IO or seeking weirdly?)
• First pass is performed after 51 frames have been written to the output according to the console status line
• This behavior is only present when using -f concat and not when processing a single file
• Each mp4 input file is processed twice
• The concat text file can contain different container (mixed mkv and mp4 for example) and only the MP4 files are processed twice
• The issue is present regardless of file location (Tested: local disk and SMB share)
• This issue almost doubles encoding time

Version Info

ffmpeg version N-91266-g8c20ea8ee0 Copyright (c) 2000-2018 the FFmpeg developers
built with gcc 7.3.0 (GCC)
configuration: --disable-static --enable-shared --enable-gpl --enable-version3
--enable-sdl2 --enable-bzlib --enable-fontconfig --enable-gnutls --enable-iconv
--enable-libass --enable-libbluray --enable-libfreetype --enable-libmp3lame
--enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg
--enable-libopus --enable-libshine --enable-libsnappy --enable-libsoxr
--enable-libtheora --enable-libtwolame --enable-libvpx --enable-libwavpack
--enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2
--enable-libzimg --enable-lzma --enable-zlib --enable-gmp --enable-libvidstab
--enable-libvorbis --enable-libvo-amrwbenc --enable-libmysofa --enable-libspeex
--enable-libxvid --enable-libaom --enable-libmfx --enable-amf --enable-ffnvcodec
--enable-cuvid --enable-d3d11va --enable-nvenc --enable-nvdec --enable-dxva2
--enable-avisynth
libavutil      56. 18.102 / 56. 18.102
libavcodec     58. 20.101 / 58. 20.101
libavformat    58. 17.100 / 58. 17.100
libavdevice    58.  4.100 / 58.  4.100
libavfilter     7. 25.100 /  7. 25.100
libswscale      5.  2.100 /  5.  2.100
libswresample   3.  2.100 /  3.  2.100
libpostproc    55.  2.100 / 55.  2.100

[AyrA]

compressed ffmpeg log using -report

[Carl Eugen Hoyos]

Where should we put the information that compressing the log is (extremely) counter-productive?

[AyrA]

Where should I put the info that the log you request me to add (https://ffmpeg.org/bugreports.html) does not fit the size constraints of this bug tracker? Increase the limit and I upload the uncompressed log. It's 39'031'691 bytes.

[Carl Eugen Hoyos]

Hard to believe that the issue is not reproducible with a shorter encode.

How did you create the input file?

[AyrA]

The files come from a GoPro camera. The camera splits them into 4GB chunks during creation and there is no way of avoiding it.

[Carl Eugen Hoyos]

Is the issue reproducible if you convert the input files with qt-faststart before feeding them to ffmpeg?

[AyrA]

Yes. It still stalls for each media file.

What did help though was doing ffmpeg.exe -i gopro_file.mp4 -c copy output_file.mp4 for each individual file and then concatenating these files instead. Not sure why but it looks like the GoPro writes media files in a way that messes up -f concat but works fine for individual files.

[Carl Eugen Hoyos]

Please provide a sample.

[AyrA]

​Video Parts

[Carl Eugen Hoyos]

The issue can now be shown with loglevel verbose, I don't know if this is a bug that can be fixed or an unavoidable limitation of the concat demuxer.

$ ffmpeg -loglevel verbose -f concat -i concat.txt -t 1 -f null -
ffmpeg version N-91317-gb41b6b3 Copyright (c) 2000-2018 the FFmpeg developers
  built with gcc 6.4.0 (GCC)
  configuration: --enable-gpl
  libavutil      56. 18.102 / 56. 18.102
  libavcodec     58. 20.102 / 58. 20.102
  libavformat    58. 17.100 / 58. 17.100
  libavdevice    58.  4.101 / 58.  4.101
  libavfilter     7. 25.100 /  7. 25.100
  libswscale      5.  2.100 /  5.  2.100
  libswresample   3.  2.100 /  3.  2.100
  libpostproc    55.  2.100 / 55.  2.100
[h264 @ 0x3ee14c0] Reinit context to 1920x1088, pix_fmt: yuvj420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x3edf880] Auto-inserting h264_mp4toannexb bitstream filter
[h264 @ 0x3ee52c0] Reinit context to 1920x1088, pix_fmt: yuvj420p
[concat @ 0x3ed6500] Could not find codec parameters for stream 2 (Unknown: none): unknown codec
Consider increasing the value for the 'analyzeduration' and 'probesize' options
[concat @ 0x3ed6500] Could not find codec parameters for stream 3 (Unknown: none): unknown codec
Consider increasing the value for the 'analyzeduration' and 'probesize' options
[concat @ 0x3ed6500] Could not find codec parameters for stream 4 (Unknown: none): unknown codec
Consider increasing the value for the 'analyzeduration' and 'probesize' options
Input #0, concat, from 'concat.txt':
  Duration: N/A, start: 0.000000, bitrate: 47190 kb/s
    Stream #0:0(eng): Video: h264 (High), 1 reference frame (avc1 / 0x31637661), yuvj420p(pc, bt709, left), 1920x1080 (1920x1088) [SAR 1:1 DAR 16:9], 47001 kb/s, 50 fps, 50 tbr, 90k tbn, 100 tbc
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro AVC
      encoder         : GoPro AVC encoder
      timecode        : 18:00:58:12
    Stream #0:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 189 kb/s
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro AAC
      timecode        : 18:00:58:12
    Stream #0:2: Unknown: none
    Stream #0:3: Unknown: none
    Stream #0:4: Unknown: none
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> wrapped_avframe (native))
  Stream #0:1 -> #0:1 (aac (native) -> pcm_s16le (native))
Press [q] to stop, [?] for help
[h264 @ 0x4b55840] Reinit context to 1920x1088, pix_fmt: yuvj420p
[graph_1_in_0_1 @ 0x3f18ec0] tb:1/48000 samplefmt:fltp samplerate:48000 chlayout:0x3
[format_out_0_1 @ 0x3f19540] auto-inserting filter 'auto_resampler_0' between the filter 'Parsed_anull_0' and the filter 'format_out_0_1'
[auto_resampler_0 @ 0x3f1a580] ch:2 chl:stereo fmt:fltp r:48000Hz -> ch:2 chl:stereo fmt:s16 r:48000Hz
[graph 0 input from stream 0:0 @ 0x3f2fd80] w:1920 h:1080 pixfmt:yuvj420p tb:1/90000 fr:50/1 sar:1/1 sws_param:flags=2
Output #0, null, to 'pipe:':
  Metadata:
    encoder         : Lavf58.17.100
    Stream #0:0(eng): Video: wrapped_avframe, 1 reference frame, yuvj420p(left), 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 200 kb/s, 50 fps, 50 tbn, 50 tbc
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro AVC
      timecode        : 18:00:58:12
      encoder         : Lavc58.20.102 wrapped_avframe
    Stream #0:1(eng): Audio: pcm_s16le, 48000 Hz, stereo, s16, 1536 kb/s
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro AAC
      timecode        : 18:00:58:12
      encoder         : Lavc58.20.102 pcm_s16le
No more output streams to write to, finishing.:01.00 bitrate=N/A speed=0.0226x
frame=   50 fps=1.1 q=-0.0 Lsize=N/A time=00:00:01.00 bitrate=N/A speed=0.0224x
video:26kB audio:188kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown
Input file #0 (concat.txt):
  Input stream #0:0 (video): 59 packets read (7258062 bytes); 52 frames decoded;
  Input stream #0:1 (audio): 55 packets read (27760 bytes); 55 frames decoded (56320 samples);
  Input stream #0:2 ((null)): 1 packets read (4 bytes);
  Input stream #0:3 ((null)): 1 packets read (4428 bytes);
  Input stream #0:4 ((null)): 66362 packets read (1062472 bytes);
  Total: 66478 packets (8352726 bytes) demuxed
Output file #0 (pipe:):
  Output stream #0:0 (video): 50 frames encoded; 50 packets muxed (26800 bytes);
  Output stream #0:1 (audio): 47 frames encoded (48000 samples); 47 packets muxed (192000 bytes);
  Total: 97 packets (218800 bytes) muxed
[AVIOContext @ 0x3ee8100] Statistics: 1117648442 bytes read, 33854 seeks
[AVIOContext @ 0x3ed6c80] Statistics: 19 bytes read, 0 seeks

$ ffmpeg -loglevel verbose -i GH010051.MP4 -t 1 -f null -
ffmpeg version N-91317-gb41b6b3 Copyright (c) 2000-2018 the FFmpeg developers
  built with gcc 6.4.0 (GCC)
  configuration: --enable-gpl
  libavutil      56. 18.102 / 56. 18.102
  libavcodec     58. 20.102 / 58. 20.102
  libavformat    58. 17.100 / 58. 17.100
  libavdevice    58.  4.101 / 58.  4.101
  libavfilter     7. 25.100 /  7. 25.100
  libswscale      5.  2.100 /  5.  2.100
  libswresample   3.  2.100 /  3.  2.100
  libpostproc    55.  2.100 / 55.  2.100
[h264 @ 0x3c00700] Reinit context to 1920x1088, pix_fmt: yuvj420p
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'GH010051.MP4':
  Metadata:
    major_brand     : mp41
    minor_version   : 538120216
    compatible_brands: mp41
    creation_time   : 2018-06-13T18:00:58.000000Z
    firmware        : HD6.01.02.01.00
  Duration: 00:11:18.02, start: 0.000000, bitrate: 47250 kb/s
    Stream #0:0(eng): Video: h264 (High), 1 reference frame (avc1 / 0x31637661), yuvj420p(pc, bt709, left), 1920x1080 (1920x1088) [SAR 1:1 DAR 16:9], 47001 kb/s, 50 fps, 50 tbr, 90k tbn, 100 tbc (default)
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro AVC
      encoder         : GoPro AVC encoder
      timecode        : 18:00:58:12
    Stream #0:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 189 kb/s (default)
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro AAC
      timecode        : 18:00:58:12
    Stream #0:2(eng): Data: none (tmcd / 0x64636D74) (default)
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro TCD
      timecode        : 18:00:58:12
    Stream #0:3(eng): Data: none (gpmd / 0x646D7067), 35 kb/s (default)
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro MET
    Stream #0:4(eng): Data: none (fdsc / 0x63736466), 12 kb/s (default)
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro SOS
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> wrapped_avframe (native))
  Stream #0:1 -> #0:1 (aac (native) -> pcm_s16le (native))
Press [q] to stop, [?] for help
[h264 @ 0x3c273c0] Reinit context to 1920x1088, pix_fmt: yuvj420p
[graph_1_in_0_1 @ 0x3eb04c0] tb:1/48000 samplefmt:fltp samplerate:48000 chlayout:0x3
[format_out_0_1 @ 0x3eb0e40] auto-inserting filter 'auto_resampler_0' between the filter 'Parsed_anull_0' and the filter 'format_out_0_1'
[auto_resampler_0 @ 0x3eb2340] ch:2 chl:stereo fmt:fltp r:48000Hz -> ch:2 chl:stereo fmt:s16 r:48000Hz
[graph 0 input from stream 0:0 @ 0x4665580] w:1920 h:1080 pixfmt:yuvj420p tb:1/90000 fr:50/1 sar:1/1 sws_param:flags=2
Output #0, null, to 'pipe:':
  Metadata:
    major_brand     : mp41
    minor_version   : 538120216
    compatible_brands: mp41
    firmware        : HD6.01.02.01.00
    encoder         : Lavf58.17.100
    Stream #0:0(eng): Video: wrapped_avframe, 1 reference frame, yuvj420p(left), 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 200 kb/s, 50 fps, 50 tbn, 50 tbc (default)
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro AVC
      timecode        : 18:00:58:12
      encoder         : Lavc58.20.102 wrapped_avframe
    Stream #0:1(eng): Audio: pcm_s16le, 48000 Hz, stereo, s16, 1536 kb/s (default)
    Metadata:
      creation_time   : 2018-06-13T18:00:58.000000Z
      handler_name    : GoPro AAC
      timecode        : 18:00:58:12
      encoder         : Lavc58.20.102 pcm_s16le
No more output streams to write to, finishing.
frame=   50 fps=0.0 q=-0.0 Lsize=N/A time=00:00:01.00 bitrate=N/A speed=4.76x
video:26kB audio:188kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown
Input file #0 (GH010051.MP4):
  Input stream #0:0 (video): 59 packets read (7257987 bytes); 52 frames decoded;
  Input stream #0:1 (audio): 55 packets read (27760 bytes); 55 frames decoded (56320 samples);
  Input stream #0:2 (data): 1 packets read (4 bytes);
  Input stream #0:3 (data): 0 packets read (0 bytes);
  Input stream #0:4 (data): 4 packets read (744 bytes);
  Total: 119 packets (7286495 bytes) demuxed
Output file #0 (pipe:):
  Output stream #0:0 (video): 50 frames encoded; 50 packets muxed (26800 bytes);
  Output stream #0:1 (audio): 47 frames encoded (48000 samples); 47 packets muxed (192000 bytes);
  Total: 97 packets (218800 bytes) muxed
[AVIOContext @ 0x3c07180] Statistics: 8451658 bytes read, 4 seeks

The seeks/reads take considerable time, concat.txt contains only the line file GH010051.MP4, remuxing the input file with FFmpeg fixes the issue.