mov regression

[Carl Eugen Hoyos]

​http://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket2605/
The samples from ticket #2605 are not transcoded in-sync since ca6cae73

$ ffmpeg -i audio_silence_after_ffmpeg_upgrade.mov -qscale 2 out.avi
ffmpeg version N-82095-g5867234 Copyright (c) 2000-2016 the FFmpeg developers
  built with gcc 4.7 (SUSE Linux)
  configuration: --enable-gpl
  libavutil      55. 33.100 / 55. 33.100
  libavcodec     57. 63.103 / 57. 63.103
  libavformat    57. 53.100 / 57. 53.100
  libavdevice    57.  0.103 / 57.  0.103
  libavfilter     6. 64.100 /  6. 64.100
  libswscale      4.  1.100 /  4.  1.100
  libswresample   2.  2.100 /  2.  2.100
  libpostproc    54.  0.100 / 54.  0.100
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x3a583c0] STSZ sample size 1 invalid (too small), ignoring
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'audio_silence_after_ffmpeg_upgrade.mov':
  Metadata:
    major_brand     : qt
    minor_version   : 512
    compatible_brands: qt
    encoder         : Lavf55.7.100
  Duration: 00:00:15.60, start: 0.000000, bitrate: 1768 kb/s
    Stream #0:0(eng): Video: h264 (Baseline) (avc1 / 0x31637661), yuv420p, 480x270 [SAR 1:1 DAR 16:9], 361 kb/s, 25 fps, 25 tbr, 25k tbn, 50k tbc (default)
    Metadata:
      handler_name    : DataHandler
    Stream #0:1(eng): Audio: pcm_s16le (sowt / 0x74776F73), 44100 Hz, stereo, s16, 1411 kb/s (default)
    Metadata:
      handler_name    : DataHandler
Please use -q:a or -q:v, -qscale is ambiguous
Output #0, avi, to 'out.avi':
  Metadata:
    major_brand     : qt
    minor_version   : 512
    compatible_brands: qt
    ISFT            : Lavf57.53.100
    Stream #0:0(eng): Video: mpeg4 (FMP4 / 0x34504D46), yuv420p, 480x270 [SAR 1:1 DAR 16:9], q=2-31, 200 kb/s, 25 fps, 25 tbn, 25 tbc (default)
    Metadata:
      handler_name    : DataHandler
      encoder         : Lavc57.63.103 mpeg4
    Side data:
      cpb: bitrate max/min/avg: 0/0/200000 buffer size: 0 vbv_delay: -1
    Stream #0:1(eng): Audio: ac3 ([0] [0][0] / 0x2000), 44100 Hz, stereo, fltp, 192 kb/s (default)
    Metadata:
      handler_name    : DataHandler
      encoder         : Lavc57.63.103 ac3
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> mpeg4 (native))
  Stream #0:1 -> #0:1 (pcm_s16le (native) -> ac3 (native))
Press [q] to stop, [?] for help
[avi @ 0x3c336e0] Non-monotonous DTS in output stream 0:1; previous: 2, current: 0; changing to 3. This may result in incorrect timestamps in the output file.
[avi @ 0x3c336e0] Non-monotonous DTS in output stream 0:1; previous: 3, current: 0; changing to 4. This may result in incorrect timestamps in the output file.
[avi @ 0x3c336e0] Non-monotonous DTS in output stream 0:1; previous: 4, current: 0; changing to 5. This may result in incorrect timestamps in the output file.

...

...

[avi @ 0x3c336e0] Non-monotonous DTS in output stream 0:1; previous: 221, current: 0; changing to 222. This may result in incorrect timestamps in the output file.
[avi @ 0x3c336e0] Non-monotonous DTS in output stream 0:1; previous: 222, current: 0; changing to 223. This may result in incorrect timestamps in the output file.
[avi @ 0x3c336e0] Non-monotonous DTS in output stream 0:1; previous: 223, current: 0; changing to 224. This may result in incorrect timestamps in the output file.
frame=  381 fps=150 q=2.0 Lsize=     933kB time=00:00:15.36 bitrate= 497.6kbits/s speed=6.04x 
video:725kB audio:184kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 2.664956%

[Sasi Inguva]

stts:
sample count , sample duration
688128 , 0
0, 0

The sample duration of in stts atom is zero . I don't know if we need to support such file?
The problem is that there are 688128 audio samples with zero timestamp in the audio stream index. With edit list patch we are seeking to timestamp zero for the editlist entry, using av_index_search_timestamp which does a binary search and lands somewhere in the middle. So half the audio samples are removed from the index. Its easy to fix the search function, but don't know if we need to support such file which seems to be corrupt ?

[Carl Eugen Hoyos]

Replying to isasi:

stts:
sample count , sample duration
688128 , 0
0, 0

The sample duration of in stts atom is zero . I don't know if we need to support such file?

It played fine by default in the past.

The problem is that there are 688128 audio samples with zero timestamp in the audio stream index. With edit list patch we are seeking to timestamp zero for the editlist entry, using av_index_search_timestamp which does a binary search and lands somewhere in the middle. So half the audio samples are removed from the index.

This sounds as if it would be possible to fix this specific sample.

[Sasi Inguva]

patch sent.

[Carl Eugen Hoyos]

Fixed by Sasi Inguva in 7e538c947547b04267f0b307b0e7d96a84d558e0