[FFmpeg-devel] Captions SCC

Devin Heitmueller devin.heitmueller at ltnglobal.com
Sat Feb 8 01:12:10 EET 2025


On Fri, Feb 7, 2025 at 5:18 PM Marth64 <marth64 at proxyid.net> wrote:
> (libavcodec’s definition of AV_CODEC_ID_EIA_608 is a fallacy, it is
> actually A/53 Part 4 coding that is passed around with this codec ID,
> which can contain multiple sub-streams including EIA-608 and CEA-708)

I probably wouldn't call it a "fallacy".  I would say it's highly
inconsistent.  In some cases it contains 608 with both pairs, in some
cases with only one pair, in some cases it contains both 608 and 708.
 Whether it works properly in all those cases varies wildly depending
on which muxer and demuxers are involved (there are lots of cases
where the 708 gets stripped out unnecessarily, or the format only
assumes 608 will ever be present and thus improperly handles the 708
data).

It is a huge mess.  I suspect to get things into more reasonable shape
would require a set of test vectors that exercise all the different
combinations of muxer/demuxer/encoder/decoders that support captions.

Devin

-- 
Devin Heitmueller, Senior Software Engineer
LTN Global Communications
o: +1 (301) 363-1001
w: https://ltnglobal.com  e: devin.heitmueller at ltnglobal.com


More information about the ffmpeg-devel mailing list