Re: [Cellar] Matroska support for closed captions?

[Jerome Martinez <[e-mail]>]

On 27/04/2020 17:46, Dave Rice wrote:
>
> For EIA-608 and EIA-708 data I suppose, either a new type of Codec Mapping could be written (and then these would be a new type of subtitle track) or they could be support as side-data with a BlockAdditionalMapping as described at https://www.ietf.org/id/draft-ietf-cellar-codec-04.html#name-block-additional-mapping. I’m unsure which technique the community would recommend but curious for some discussion on it.


As MXF have JPEG-2000, I guess that captions are in a standalone 
ancillary data (SMPTE ST 291) track.
As a first shot, I would not be in favor of BlockAdditionalMapping 
because the track does not depend on data in another bistream.

then there would (at least) 2 choices:
- demux of the track from MXF and mux of this track as is; advantage is 
that we lose no other data inside the ancillary data, disadvantage is 
that lot of different content (several formats of captions, bar data, 
VBI, time codes, Acquisition Metadata...) are muxed in the "raw" stream, 
more difficult for a demuxer/player.
- demux of the captions from Ancillary data from MXF; advantage is that 
we have 1 track per content stream, disadvantage is that we may lose 
some other "opaque" data from the ancillary data.

That said, we could propose both.

Jérôme