copy via HTTP with no transcode very slow for MOV vs. MP4

[Daniel Heidebrecht]

Summary of the bug: Copying a section from an MOV file with no transcoding via HTTP is very slow compared to the same operation on an MP4 file. In the example below - MOV elapsed time is 11 minutes, MP4 elapsed time is ~ 8 seconds.

How to reproduce: The original test file is an 85Mb MOV file, the MP4 comparison file was created by converting the MOV file to MP4:

ffmpeg -i test.mov -c copy test-copy.mp4

The MOV and MP4 files are both publicly available on S3:

​https://s3.amazonaws.com/unclipvideo-test/test.mov
​https://s3.amazonaws.com/unclipvideo-test/test-copy.mp4

MOV and MP4 example test output:

MOV test

time ./ffmpeg -ss 5 -i https://s3.amazonaws.com/unclipvideo-test/test.mov -t 15 -c copy test-s3.mov
ffmpeg version N-81504-g1fb8f6b-tessus Copyright (c) 2000-2016 the FFmpeg developers
  built with Apple LLVM version 6.0 (clang-600.0.57) (based on LLVM 3.5svn)
  configuration: --cc=/usr/bin/clang --prefix=/opt/ffmpeg --as=yasm --extra-version=tessus --enable-avisynth --enable-fontconfig --enable-gpl --enable-libass --enable-libbluray --enable-libfreetype --enable-libgsm --enable-libmodplug --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopus --enable-libschroedinger --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libx264 --enable-libx265 --enable-libxavs --enable-libxvid --enable-libzmq --enable-version3 --disable-ffplay --disable-indev=qtkit --disable-indev=x11grab_xcb
  libavutil      55. 28.100 / 55. 28.100
  libavcodec     57. 51.100 / 57. 51.100
  libavformat    57. 46.100 / 57. 46.100
  libavdevice    57.  0.102 / 57.  0.102
  libavfilter     6. 49.100 /  6. 49.100
  libswscale      4.  1.100 /  4.  1.100
  libswresample   2.  1.100 /  2.  1.100
  libpostproc    54.  0.100 / 54.  0.100
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'https://s3.amazonaws.com/unclipvideo-test/test.mov':
  Metadata:
    major_brand     : qt  
    minor_version   : 0
    compatible_brands: qt  
    creation_time   : 2016-07-31 22:04:54
    com.apple.quicktime.location.ISO6709: +41.3207-071.9074-000.385/
    com.apple.quicktime.make: Apple
    com.apple.quicktime.model: iPhone 6s
    com.apple.quicktime.software: 9.3.3
    com.apple.quicktime.creationdate: 2016-07-31T18:04:54-0400
  Duration: 00:00:40.80, start: 0.000000, bitrate: 17134 kb/s
    Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709), 1920x1080, 17042 kb/s, 29.98 fps, 29.97 tbr, 600 tbn, 1200 tbc (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
      encoder         : H.264
    Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, mono, fltp, 82 kb/s (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
    Stream #0:2(und): Data: none (mebx / 0x7862656D), 0 kb/s (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
    Stream #0:3(und): Data: none (mebx / 0x7862656D), 0 kb/s (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
[mov @ 0x7fe6f3002e00] Using AVStream.codec to pass codec parameters to muxers is deprecated, use AVStream.codecpar instead.
    Last message repeated 1 times
Output #0, mov, to 'test-s3.mov':
  Metadata:
    major_brand     : qt  
    minor_version   : 0
    compatible_brands: qt  
    com.apple.quicktime.creationdate: 2016-07-31T18:04:54-0400
    com.apple.quicktime.location.ISO6709: +41.3207-071.9074-000.385/
    com.apple.quicktime.make: Apple
    com.apple.quicktime.model: iPhone 6s
    com.apple.quicktime.software: 9.3.3
    encoder         : Lavf57.46.100
    Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709), 1920x1080, q=2-31, 17042 kb/s, 29.98 fps, 29.97 tbr, 19200 tbn, 600 tbc (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
      encoder         : H.264
    Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, mono, 82 kb/s (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
Stream mapping:
  Stream #0:0 -> #0:0 (copy)
  Stream #0:1 -> #0:1 (copy)
Press [q] to stop, [?] for help
frame=  480 fps=0.7 q=-1.0 Lsize=   33291kB time=00:00:15.01 bitrate=18162.3kbits/s speed=0.0227x    
video:33113kB audio:163kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.044288%

real	11m5.040s
user	0m3.638s
sys	0m1.768s

MP4 test:

time ./ffmpeg -ss 5 -i https://s3.amazonaws.com/unclipvideo-test/test-copy.mp4 -t 15 -c copy test-s3.mp4
ffmpeg version N-81504-g1fb8f6b-tessus Copyright (c) 2000-2016 the FFmpeg developers
  built with Apple LLVM version 6.0 (clang-600.0.57) (based on LLVM 3.5svn)
  configuration: --cc=/usr/bin/clang --prefix=/opt/ffmpeg --as=yasm --extra-version=tessus --enable-avisynth --enable-fontconfig --enable-gpl --enable-libass --enable-libbluray --enable-libfreetype --enable-libgsm --enable-libmodplug --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopus --enable-libschroedinger --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libx264 --enable-libx265 --enable-libxavs --enable-libxvid --enable-libzmq --enable-version3 --disable-ffplay --disable-indev=qtkit --disable-indev=x11grab_xcb
  libavutil      55. 28.100 / 55. 28.100
  libavcodec     57. 51.100 / 57. 51.100
  libavformat    57. 46.100 / 57. 46.100
  libavdevice    57.  0.102 / 57.  0.102
  libavfilter     6. 49.100 /  6. 49.100
  libswscale      4.  1.100 /  4.  1.100
  libswresample   2.  1.100 /  2.  1.100
  libpostproc    54.  0.100 / 54.  0.100
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'https://s3.amazonaws.com/unclipvideo-test/test-copy.mp4':
  Metadata:
    major_brand     : isom
    minor_version   : 512
    compatible_brands: isomiso2avc1mp41
    encoder         : Lavf57.25.100
  Duration: 00:00:40.87, start: 0.000000, bitrate: 17102 kb/s
    Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709), 1920x1080, 17042 kb/s, 29.98 fps, 29.97 tbr, 19200 tbn, 38400 tbc (default)
    Metadata:
      handler_name    : VideoHandler
    Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, mono, fltp, 82 kb/s (default)
    Metadata:
      handler_name    : SoundHandler
[mp4 @ 0x7fcc5901ac00] Using AVStream.codec to pass codec parameters to muxers is deprecated, use AVStream.codecpar instead.
    Last message repeated 1 times
Output #0, mp4, to 'test-s3.mp4':
  Metadata:
    major_brand     : isom
    minor_version   : 512
    compatible_brands: isomiso2avc1mp41
    encoder         : Lavf57.46.100
    Stream #0:0(und): Video: h264 (High) ([33][0][0][0] / 0x0021), yuv420p(tv, bt709), 1920x1080, q=2-31, 17042 kb/s, 29.98 fps, 29.97 tbr, 19200 tbn, 19200 tbc (default)
    Metadata:
      handler_name    : VideoHandler
    Stream #0:1(und): Audio: aac (LC) ([64][0][0][0] / 0x0040), 44100 Hz, mono, 82 kb/s (default)
    Metadata:
      handler_name    : SoundHandler
Stream mapping:
  Stream #0:0 -> #0:0 (copy)
  Stream #0:1 -> #0:1 (copy)
Press [q] to stop, [?] for help
frame=  480 fps=104 q=-1.0 Lsize=   33291kB time=00:00:15.01 bitrate=18162.2kbits/s speed=3.25x    
video:33113kB audio:163kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.044021%

real	0m7.502s
user	0m0.290s
sys	0m0.203s

ffmpeg version:

ffmpeg version N-81504-g1fb8f6b-tessus Copyright (c) 2000-2016 the FFmpeg developers
built with Apple LLVM version 6.0 (clang-600.0.57) (based on LLVM 3.5svn)

[Carl Eugen Hoyos]

The fact that FFmpeg seeks wildly in the original mov file (but not in a file remuxed with FFmpeg) leading to many slow reconnections is reproducible with the following command line:

$ ffmpeg -loglevel debug -i test.mov -c copy -f null -
ffmpeg version N-81379-gf866f22 Copyright (c) 2000-2016 the FFmpeg developers
  built with gcc 4.7 (SUSE Linux)
  configuration: --enable-gpl --enable-gnutls
  libavutil      55. 28.100 / 55. 28.100
  libavcodec     57. 53.100 / 57. 53.100
  libavformat    57. 47.101 / 57. 47.101
  libavdevice    57.  0.102 / 57.  0.102
  libavfilter     6. 52.100 /  6. 52.100
  libswscale      4.  1.100 /  4.  1.100
  libswresample   2.  1.100 /  2.  1.100
  libpostproc    54.  0.100 / 54.  0.100
Splitting the commandline.
Reading option '-loglevel' ... matched as option 'loglevel' (set logging level) with argument 'debug'.
Reading option '-i' ... matched as input file with argument 'test.mov'.
Reading option '-c' ... matched as option 'c' (codec name) with argument 'copy'.
Reading option '-f' ... matched as option 'f' (force format) with argument 'null'.
Reading option '-' ... matched as output file.
Finished splitting the commandline.
Parsing a group of options: global .
Applying option loglevel (set logging level) with argument debug.
Successfully parsed a group of options.
Parsing a group of options: input file test.mov.
Successfully parsed a group of options.
Opening an input file: test.mov.
[file @ 0x28a4e80] Setting default whitelist 'file,crypto'
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] Format mov,mp4,m4a,3gp,3g2,mj2 probed with size=2048 and score=100
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] ISO: File Type Major Brand: qt
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] Unknown dref type 0x0873696c61 size 12
    Last message repeated 3 times
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] rfps: 29.833333 0.019205
    Last message repeated 1 times
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] rfps: 29.916667 0.003699
    Last message repeated 1 times
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] rfps: 30.000000 0.000558
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] rfps: 60.000000 0.002233
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] rfps: 120.000000 0.008934
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] rfps: 29.970030 0.000264
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] rfps: 59.940060 0.001055
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] Before avformat_find_stream_info() pos: 87380080 bytes read:53138 seeks:1 nb_streams:4
[h264 @ 0x28a66a0] nal_unit_type: 7, nal_ref_idc: 1
[h264 @ 0x28a66a0] nal_unit_type: 8, nal_ref_idc: 1
[h264 @ 0x28a66a0] nal_unit_type: 6, nal_ref_idc: 0
[h264 @ 0x28a66a0] nal_unit_type: 5, nal_ref_idc: 1
[h264 @ 0x28a66a0] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] All info found
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x28a4640] After avformat_find_stream_info() pos: 206381 bytes read:259483 seeks:2 frames:45
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'test.mov':
  Metadata:
    major_brand     : qt
    minor_version   : 0
    compatible_brands: qt
    creation_time   : 2016-07-31 22:04:54
    com.apple.quicktime.location.ISO6709: +41.3207-071.9074-000.385/
    com.apple.quicktime.make: Apple
    com.apple.quicktime.model: iPhone 6s
    com.apple.quicktime.software: 9.3.3
    com.apple.quicktime.creationdate: 2016-07-31T18:04:54-0400
  Duration: 00:00:40.80, start: 0.000000, bitrate: 17134 kb/s
    Stream #0:0(und), 1, 1/600: Video: h264 (High), 1 reference frame (avc1 / 0x31637661), yuv420p(tv, bt709, left), 1920x1080 (1920x1088), 0/1, 17042 kb/s, 29.98 fps, 29.97 tbr, 600 tbn, 1200 tbc (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
      encoder         : H.264
    Stream #0:1(und), 44, 1/44100: Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, mono, fltp, 82 kb/s (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
    Stream #0:2(und), 0, 1/600: Data: none (mebx / 0x7862656D), 0/1, 0 kb/s (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
    Stream #0:3(und), 0, 1/600: Data: none (mebx / 0x7862656D), 0/1, 0 kb/s (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
Successfully opened the file.
Parsing a group of options: output file -.
Applying option c (codec name) with argument copy.
Applying option f (force format) with argument null.
Successfully parsed a group of options.
Opening an output file: -.
Successfully opened the file.
[null @ 0x28f2a60] Using AVStream.codec to pass codec parameters to muxers is deprecated, use AVStream.codecpar instead.
    Last message repeated 1 times
Output #0, null, to 'pipe:':
  Metadata:
    major_brand     : qt
    minor_version   : 0
    compatible_brands: qt
    com.apple.quicktime.creationdate: 2016-07-31T18:04:54-0400
    com.apple.quicktime.location.ISO6709: +41.3207-071.9074-000.385/
    com.apple.quicktime.make: Apple
    com.apple.quicktime.model: iPhone 6s
    com.apple.quicktime.software: 9.3.3
    encoder         : Lavf57.47.101
    Stream #0:0(und), 0, 1/600: Video: h264 (High), 1 reference frame (avc1 / 0x31637661), yuv420p(tv, bt709, left), 1920x1080 (0x0), 0/1, q=2-31, 17042 kb/s, 29.98 fps, 29.97 tbr, 600 tbn, 600 tbc (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
      encoder         : H.264
    Stream #0:1(und), 0, 1/44100: Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, mono, 82 kb/s (default)
    Metadata:
      creation_time   : 2016-07-31 22:04:54
      handler_name    : Core Media Data Handler
Stream mapping:
  Stream #0:0 -> #0:0 (copy)
  Stream #0:1 -> #0:1 (copy)
Press [q] to stop, [?] for help
cur_dts is invalid (this is harmless if it occurs once at the start per stream)
    Last message repeated 45 times
No more output streams to write to, finishing.
frame= 1223 fps=0.0 q=-1.0 Lsize=N/A time=00:00:40.86 bitrate=N/A speed=1.43e+03x
video:84872kB audio:413kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown
Input file #0 (test.mov):
  Input stream #0:0 (video): 1223 packets read (86909375 bytes);
  Input stream #0:1 (audio): 1760 packets read (423378 bytes);
  Input stream #0:2 (data): 0 packets read (0 bytes);
  Input stream #0:3 (data): 0 packets read (0 bytes);
  Total: 2983 packets (87332753 bytes) demuxed
Output file #0 (pipe:):
  Output stream #0:0 (video): 1223 packets muxed (86909375 bytes);
  Output stream #0:1 (audio): 1760 packets muxed (423378 bytes);
  Total: 2983 packets (87332753 bytes) muxed
0 frames successfully decoded, 0 decoding errors
[AVIOContext @ 0x28a5160] Statistics: 127025736 bytes read, 2419 seeks

I don't know if there is a bug that can be fixed in FFmpeg.

[Carl Eugen Hoyos]

All relevant seeks happen in mov_read_packet().