Timing and Persistence

[ronan-boitard]

Timing and persistence of metadata are not mentioned in the current draft. It would be desirable to get a clause mentioning how they are handled for T.35 messages

Proposal:
D.4 Timing Signalling
Various ITU T.35 carrying mechanisms exist to indicate the 'TimeIntervalStart' and 'TimeIntervalDuration' metadata. If a carrying mechanism provides a native means of persistence for ITU-T T.35 metadata, then that could be used to indicate whether the signaled SMPTE ST 2094-50 metadata applies to one or more frames. However, if such a mechanism does not exist then the metadata shall be included in every frame that requires such information.

[ccameron-chromium]

For content where the metadata does not change every frame, the each-frame-repetition can add up.

The default behavior that I would suggest for ITU-T T.35 metadata would be to have have it persist. If need be, we can add a "do not persist" flag.