Re: [imsc] Maximum number of regions

Apologies, this was sent to the wrong group - it was intended to go to
TTWG. 


On 12/10/2015 09:31, "Nigel Megitt" <nigel.megitt@bbc.co.uk> wrote:

>All,
>
>I remain concerned about the limit of maximum number of simultaneously
>active regions in IMSC 1.
>
>A colleague has scanned the BBC's archive of subtitles and discovered one
>file that would require 4 regions, and about 0.01% of files require 3
>regions.
>
>I attach a screenshot from one of the 3 region files. It is admittedly
>rather old and not representative of present day practices however when we
>repeat the programme we reuse those original subtitles.
>
>An additional use case for more regions is when the subtitle file contains
>an accessible version of on-screen text ("object based broadcasting"), for
>example a combination of the hard of hearing or translation subtitles,
>clock, news ticker, "Breaking News" caption, contributor name and job
>title etc., as currently visible on rolling news channels e.g. BBC News.
>In that scenario it would be easy to exceed the 4 region limit - I got to
>5 or 6 just then without any special cases. Moving to this method of
>presentation is in my opinion both an area of research and a direction of
>travel.
>
>My conclusion from looking at this is that although the current incidence
>of subtitles needing 4 regions is extremely low, at least in UK English
>language BBC content, the "safety margin" present with a 4 region limit is
>too small and it is likely that a 4 region limit on IMSC is too
>constraining of future possible uses and presentation styles.
>
>My preferred proposal for addressing this is simply to remove the
>simultaneous region limit altogether, since I do not believe there's a
>strong implementation case for having it, especially when rendering
>complexity is already tackled by the HRM.
>
>My less preferred proposal, if there is a genuine reason for needing such
>a limit, is to increase the value from 4 to 10, to allow for some
>additional safety margin.
>
>Kind regards,
>
>Nigel
>
>
>
>
>
>
>
>

Received on Monday, 12 October 2015 08:35:34 UTC