On the AOMedia ITU-T T.35 metadata syntax

[podborski]

You should not re-define the metadata_itut_t35. Instead, remove it and just specify the metadata_aom_itut_t35 like this:

metadata_aom_itut_t35( ) {	Type
itu_t_t35_country_code = 0xB5	f(8)
itu_t_t35_terminal_provider_code = 0x5890	f(16)
itu_t_t35_terminal_provider_oriented_code	f(16)
if ( itu_t_t35_terminal_provider_oriented_code == 0x0001 ) {
av1_film_grain_param_sets( )
}
}

This would combine what you currently have in clause 5.2 and 5.3.

[podborski]

The following note is not needed if the above syntax is accepted. It is already obvious that this is the AOM T.35 from the proposed syntax above. The text from the note can move as generic description in the associated semantics clause.

Note: The syntax above represents signaling of ITU-T T.35 metadata reserved for AOMedia and not general ITU-T T.35 syntax.

[podborski]

@cconcolato what do you think? The idea would be to move this one out to the AOM Metadata Registry spec.

[cconcolato]

@podborski Initially I thought there should be 3 specs:

• The film grain spec defining only the syntax of the film grain payload. This would be a regular Deliverable of the Codec WG with its usual process.
• A new spec defining the payload of the AOM T.35 metadata obu syntax, i.e. defining only on how many bits the terminal_provider_oriented_code would be. This would be a regular Deliverable of the STF WG with its usual process.
• The metadata registry that would associate the value 1 of terminal_provider_oriented_code with the film grain spec. This would not be a regular Deliverable in the sense that it would not need to go through WD/WGA WD/FD ... but more lightweight as it would only be a registry (similar to W3C new registry process).

Since we are time-constrained, I'm fine with the Codec WG going ahead here and then we can do editorial work to split the spec and transfer ownership.

[podborski]

@cconcolato I feel like 2 and 3 can be combined in one spec.