Add bitstream filter to remove Closed Captions from h264

[Carl Eugen Hoyos]

​http://thread.gmane.org/gmane.comp.video.ffmpeg.user/61438/focus=61440
The sample from ticket #2885 contains Closed Captions, FFmpeg does not allow to remove them without reencoding.

$ ffmpeg -i transformers.ts -vcodec copy out.h264
ffmpeg version N-78869-gdd2ea5c Copyright (c) 2000-2016 the FFmpeg developers
  built with gcc 4.7 (SUSE Linux)
  configuration: --enable-gpl
  libavutil      55. 19.100 / 55. 19.100
  libavcodec     57. 27.101 / 57. 27.101
  libavformat    57. 27.102 / 57. 27.102
  libavdevice    57.  0.101 / 57.  0.101
  libavfilter     6. 38.100 /  6. 38.100
  libswscale      4.  0.100 /  4.  0.100
  libswresample   2.  0.101 /  2.  0.101
  libpostproc    54.  0.100 / 54.  0.100
Input #0, mpegts, from 'transformers.ts':
  Duration: 00:01:59.29, start: 0.578456, bitrate: 9015 kb/s
  Program 1
    Stream #0:0[0x1e1]: Video: h264 (High) ([27][0][0][0] / 0x001B), yuvj420p(pc), 1920x1080 [SAR 1:1 DAR 16:9], Closed Captions, 29.97 fps, 29.97 tbr, 90k tbn, 59.94 tbc
    Stream #0:1[0x1e2]: Audio: ac3 ([129][0][0][0] / 0x0081), 48000 Hz, 5.1(side), fltp, 384 kb/s
Output #0, h264, to 'out.h264':
  Metadata:
    encoder         : Lavf57.27.102
    Stream #0:0: Video: h264 ([27][0][0][0] / 0x001B), yuvj420p, 1920x1080 [SAR 1:1 DAR 16:9], q=2-31, 29.97 fps, 29.97 tbr, 29.97 tbn, 29.97 tbc
Stream mapping:
  Stream #0:0 -> #0:0 (copy)
Press [q] to stop, [?] for help
frame= 3575 fps=0.0 q=-1.0 Lsize=  118242kB time=00:01:59.25 bitrate=8122.6kbits/s speed= 580x
video:118291kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown

$ ffmpeg -i out.h264
ffmpeg version N-78869-gdd2ea5c Copyright (c) 2000-2016 the FFmpeg developers
  built with gcc 4.7 (SUSE Linux)
  configuration: --enable-gpl
  libavutil      55. 19.100 / 55. 19.100
  libavcodec     57. 27.101 / 57. 27.101
  libavformat    57. 27.102 / 57. 27.102
  libavdevice    57.  0.101 / 57.  0.101
  libavfilter     6. 38.100 /  6. 38.100
  libswscale      4.  0.100 /  4.  0.100
  libswresample   2.  0.101 /  2.  0.101
  libpostproc    54.  0.100 / 54.  0.100
[h264 @ 0x3a8c280] Stream #0: not enough frames to estimate rate; consider increasing probesize
Input #0, h264, from 'out.h264':
  Duration: N/A, bitrate: N/A
    Stream #0:0: Video: h264 (High), yuvj420p(pc), 1920x1080 [SAR 1:1 DAR 16:9], Closed Captions, 29.97 fps, 29.97 tbr, 1200k tbn, 59.94 tbc
At least one output file must be specified

[feno]

I'm very surprised to see that this feature does not exist in ffmpeg. I was almost sure I could do that with ffmpeg so I was looking to docs to see how to do that. Then found out it's actually not possible.

I found link to this issue in some forums where other people were asking how to do this in ffmpeg. So it's something that people actually need and ask around.

No response to this issue for years since it was reported. So looks like no hope to see it implemented :(

I really need this function so badly, would appreciate a lot if any dev interested in implementing it :\

[Balling]

It is actually possible, just like with primaries/trc/matrix OF THE SOURCE that is in prefix type (39 for HEVC) of SEI.

FFmpeg has a bitstream filter called filter_units, which allows you to pass or reject NAL units by type. So we can use this to remove all the SEI NAL units, which strips out the captions.

ffmpeg -i input.mkv -c copy -bsf:v "filter_units=remove_types=6" output.mkv

See: ​https://stackoverflow.com/questions/48177694/removing-eia-608-closed-captions-from-h-264-without-reencode

Now, that is not perfect, since there may be other stuff in SEI.

Indeed, ffmpeg -i transformers_EIA608_H264.h264 -c copy -bsf:v trace_headers -f null -

will show last_payload_type_byte number 4 (that means ITU_T_T35) and itu_t_t35_country_code is 181, next 2 bytes are 49 then GA94 MKTAG,

[trace_headers @ 000002128da9cb00] 48          itu_t_t35_payload_byte[3]                            01000111 = 71
[trace_headers @ 000002128da9cb00] 56          itu_t_t35_payload_byte[4]                            01000001 = 65
[trace_headers @ 000002128da9cb00] 64          itu_t_t35_payload_byte[5]                            00111001 = 57
[trace_headers @ 000002128da9cb00] 72          itu_t_t35_payload_byte[6]                            00110100 = 52

and then 3. Just like here ​https://en.wikipedia.org/wiki/CEA-708#Picture_User_Data
​https://github.com/FFmpeg/FFmpeg/blob/870bfe16a12bf09dca3a4ae27ef6f81a2de80c40/libavcodec/sei.h#L31
I suppose the same can be done for HEVC, it uses 39 SEI though.

[Balling]

Replying to Carl Eugen Hoyos:
So you seriously want the filter that will remove all GA94 T.35 packets?

[Cigaes]

The ticket asks for a way to remove closed-captions, and nothing more. It is a perfectly valid request; there may be sensitive data in the closed captions, for example.

This ticket should not be closed until there is a way to remove exactly closed-captions.

[Balling]

Replying to Nicolas George:

The ticket asks for a way to remove closed-captions, and nothing more. It is a perfectly valid request; there may be sensitive data in the closed captions, for example.

This ticket should not be closed until there is a way to remove exactly closed-captions.

As I said there is a way to do it by removing SEI. I suppose ffmpeg should be able to remove EVEN specifically GA94 T.35 SEI, but I do not know the command (HDR10+ metadata is also stored in T.35 AFAIK, so it will remove that too, so a better narrow command is needed, I will admot as such).

[Balling]

And also may be interesting to implement this: ​https://superuser.com/questions/1356674/how-to-change-offset-of-eia-608-subtitles-in-an-mp4