Re: IMSCvNEXT + CSS styling was: TTML2 wide review comment: styling

The industry currently lacks a global subtitle standard.  IMSC1 is close
but does not support essential Japanese features.  TTML2 includes these
features, and IMSCvNext represents the minimal subset of TTML2 necessary to
achieve a global standard.  This is a project that has been in the works
for many years (Netflix has been engaged since 2014), and now have TTML2 in
WR.  We have an open-source rendering implementation
<https://github.com/skynav/ttt> for the entire feature set of TTML2 and
Netflix has built an internal workflow that covers > 50% of an
implementation.  These projects will also cover the entire feature set for
IMSCvNext.  The TTML2 work represents a significant investment of time and
money by many TTWG participants.

While Netflix is fully supportive of a CSS-based subtitle model, this this
must be TTMLvNext, or some other model.  The TTML2 train has left the
station.

Regards,
David

On Sun, Oct 1, 2017 at 3:36 PM, Silvia Pfeiffer <silviapfeiffer1@gmail.com>
wrote:

> Hi Pierre, all,
>
> With respect, in my experience, working with the CSS group to make any
> missing CSS features available is more constructive than going ahead
> and defining it separately.
>
> For example, text-wrap: balance
> (https://www.w3.org/TR/css-text-4/#text-wrap) is actually created to
> allow WebVTT to balance the different lines in a multi-line caption
> cue, which is something the caption community has asked use for. This
> may well be what you are after for multiRowAlign. But even if it's
> not, please work with the CSS group to make the features happen.
>
> Kind Regards,
> Silvia.
>
>
>
>
> On Sun, Oct 1, 2017 at 3:24 PM, Pierre-Anthony Lemieux <pal@sandflow.com>
> wrote:
> > Hi David and David et al.,
> >
> > My experience with implementing imscJS [1] has been that the mapping
> > from IMSC1 to CSS was mostly straightforward and accurate, with the
> > exception of two features that are not available natively in CSS but
> > were deemed critical to captions: linePadding and multiRowAlign.
> >
> > [1] https://github.com/sandflow/imscJS
> >
> > In my mind, IMSCvNEXT should follow a similar pattern: mapping to
> > HTML/CSS should be straightforward and accurate, with the exception of
> > features that are essential to current subtitling/captioning
> > practices, but not yet available in CSS. For such features, a fallback
> > should be specified, and UA vendors should be encourages to support
> > the feature in the future.
> >
> > Best,
> >
> > -- Pierre
> >
> > On Sat, Sep 30, 2017 at 6:01 PM, David Ronca <dronca@netflix.com> wrote:
> >>> Please consider adopting CSS as-is, without embellishment or
> improvement.
> >>
> >> CSS is beyond the scope of TTML2, and would be a requirement for
> TTMLvNext.
> >> Once of the deliverables for IMSCvNext will be a node.js  TTML->CSS
> >> transform implementation that will preserve as much of the TTML styling
> as
> >> possible.  This will simplify IMSCvNext rendering for HTML clients.
> >>
> >> On Sat, Sep 30, 2017 at 5:49 PM, David Singer <singer@apple.com> wrote:
> >>>
> >>> The styling model used in TTML2 is not CSS and is not processable by a
> >>> processor/rendering-engine designed to support HTML/CSS.  This leads to
> >>> complex ‘come from’ process deep in rendering engines, where the
> behavior
> >>> has to be dependent on whether the text ‘came from’ an HTML/CSS
> context or a
> >>> TTML context.
> >>>
> >>> Please consider adopting CSS as-is, without embellishment or
> improvement.
> >>>
> >>> David Singer
> >>> Manager, Software Standards, Apple Inc.
> >>>
> >>>
> >>
> >
>