RE: {minutes} TTWG Meeting 2015-10-01

Cyril-

TTML already has a profile parameter.  W3C itself does not need "codecs".  MPEG asked for it.  So, although we choose to construct the parameter generically, it is in practice for use by the DASH MPD.

I completely agree that the DASH MPD use of stpp is not TTML specific and therefore technically does not belong in a W3C table only about TTML profiles.

So, the question is what is the TTML media type codecs parameter? Is it for use directly by MPEG DASH MPD?  Or is it just a shorthand for the already defined profile parameter?

If the latter, then I am concerned about: 1) replication of the same thing in two parameters;  and 2) splitting the codecs definition between the W3C Note and additional guidance in MPEG (in -30 via amendment I assume).

I think we need to discuss this in MPEG.

So, for now, I mostly agree and recommend that we (you, Dave and I):

1. Remove stpp from the W3C profile registry table and proceed to a draft Note without it that W3C can liaise to MPEG;

2. Discuss the general documentation strategy of codecs values, and in particular TTML codecs values,  in 2 weeks at MPEG; and

3. Make a clarifying recommendation via liaison from MPEG to W3C about how far they should go in the media type codecs parameter registration with IANA.

Make sense?

Regards,
 Mike

-----Original Message-----
From: Cyril Concolato [mailto:cyril.concolato@telecom-paristech.fr] 
Sent: Monday, October 5, 2015 4:25 AM
To: Michael Dolan <mdolan@newtbt.com>; 'Nigel Megitt' <nigel.megitt@bbc.co.uk>; 'Glenn Adams' <glenn@skynav.com>
Cc: 'TTWG' <public-tt@w3.org>
Subject: Re: {minutes} TTWG Meeting 2015-10-01

Le 02/10/2015 17:24, Michael Dolan a écrit :
> >>  b) is partially addressed as the 'stpp' row is still there.
>
> Summarizing the last email on this – it is there because it is in use 
> today, for better or worse.  Some options: 1) delete it and thus make 
> existing use illegal; or 2) allow it but clearly deprecate its use.
>
I think you are confusing two things here:

- 'stpp' is a 4CC in the ISOBMFF. People may use 'video/mp4; codecs="stpp"' to identify a TTML track in an MP4 file. That is not wrong as TTML is an XML-based Subtitle Format, and it is currently the only standard way to signal it. Players encountering such tracks should not expect only TTML as they might problem in the future if another XML subtitle format is stored in MP4. It is MPEG's business to: a) define a more precise 'codecs' parameter for TTML-based tracks; b) inform its users to adopt the more precise 'codecs' syntax. I expect that the more precise value will be: 'video/mp4; codecs="stpp.<TTML-identifier>", where <TTML-identifier> is replaced by 'tt1f', 'tt1p', ... where the current approach 'video/mp4; codecs="stpp"' being replaced by 'video/mp4; codecs="stpp.tt1f"' (or with another identifier)

- defining 'stpp' as a TTML profile identifier is wrong, because: a) there is no "profile designator", "profile definition document", "public specification" defining the associated profile (such profile is not defined in ISOBMFF!); and b) even if it such a profile were to be defined, its use with the approach describe above, i.e. 'video/mp4; codecs="stpp.stpp"' (note the duplication of 'stpp') will be misleading!

My recommendations are:
- to TTWG: delete the 'stpp' row in the registry
- to MPEG:
     - clarify that player implementations encountering 'video/mp4; codecs="stpp" (without TTML profile identifier) should look at the file content to make sure it is TTML inside
     - define a TTML-track codecs syntax
     - recommend packagers to use the new 'codecs' syntax

HTH,
Cyril


--
Cyril Concolato
Multimedia Group / Telecom ParisTech
http://concolato.wp.mines-telecom.fr/
@cconcolato

Received on Monday, 5 October 2015 12:09:50 UTC