ogg dev - Feb 2008 - Ogg/Kate preliminary documentation

> > I did see references to Skeleton, I'll have a look at it. I
didn't
> > realize it was used widely
>
> It's not widely used currently.  The idea is to make that happen.
Oh, I get you now.
> CMML does of course other things besides subtitles.  Subtitle support
> was pretty much just added recently.  Kate however does not seem to
> offer more than CMML in the lyrics/subtitles department, besides
This is true, as the design of XML makes it malleable to custom semantics:
you can embed anything in an XML tree.
> perhaps the concept of converting other subtitles formats (ASS, SRT,
Which I have working in MPlayer now (from the 12 formats they support).
But this is, as you say, moot since it'd be as easy to create CMML
representations of these.
> This does not mean Kate development should stop.  It just means that
> if you are not willing to join efforts with the other developers we
> will end up with two (let's not count Writ) different formats for the
> same niche.  And then we have to tell people which of them should be
> the default format for subtitles and lyrics, which means one format
> will lose in favor of another, frustrating its developer(s).  I don't
> think anyone wants that.
I totally understand. I like doing this, so I will certainly not stop, but I do
see the possible counterproductive effect (which was quite apparent from
what I've seen on the web in discussions when I was looking at what was
there).
In parallel, I'll look more into CMML. However, without wanting to sound
overly dismissive, I've seen this name mentionned a few times in several
discussions over the years when text-in-ogg was mentionned, and we're
still there. Again, that's just from Google searches, so I may be missing a
lot, so please keep that in mind.

Thanks again for your feedback.

On Jan 17, 2008 2:35 AM, ogg.k.ogg.k@googlemail.com <
ogg.k.ogg.k@googlemail.com> wrote:
> > > I did see references to Skeleton, I'll have a look at it. I
didn't
> > > realize it was used widely
> >
> > It's not widely used currently.  The idea is to make that happen.
>
> Oh, I get you now.
>
> > CMML does of course other things besides subtitles.  Subtitle support
> > was pretty much just added recently.  Kate however does not seem to
> > offer more than CMML in the lyrics/subtitles department, besides
>
> This is true, as the design of XML makes it malleable to custom semantics:
> you can embed anything in an XML tree.
>
> > perhaps the concept of converting other subtitles formats (ASS, SRT,
>
> Which I have working in MPlayer now (from the 12 formats they support).
> But this is, as you say, moot since it'd be as easy to create CMML
> representations of these.
>
> > This does not mean Kate development should stop.  It just means that
> > if you are not willing to join efforts with the other developers we
> > will end up with two (let's not count Writ) different formats for
the
> > same niche.  And then we have to tell people which of them should be
> > the default format for subtitles and lyrics, which means one format
> > will lose in favor of another, frustrating its developer(s).  I
don't
> > think anyone wants that.
>
> I totally understand. I like doing this, so I will certainly not stop, but
> I do
> see the possible counterproductive effect (which was quite apparent from
> what I've seen on the web in discussions when I was looking at what was
> there).
> In parallel, I'll look more into CMML. However, without wanting to
sound
> overly dismissive, I've seen this name mentionned a few times in
several
> discussions over the years when text-in-ogg was mentionned, and we're
> still there. Again, that's just from Google searches, so I may be
missing
> a
> lot, so please keep that in mind.


There has been a lot of tool development necessary around CMML and Skeleton
in recent years, which is why CMML hasn't progressed much. Now, we are
re-designing some of CMML to address other issues that have come up. If KATE
brings out further requirements for text tracks inside Ogg, it may be easier
to include these findings into CMML to make it richer and extend those
implementations rather than have to go down that whole path again. And...
lots of elements in CMML are optional (as are the URLs for example), so if
you just want to use it for karaoke, you might be looking at a timed text
file and a CSS specification that brings out the timed colouring of the text
etc. This is the preferred way of looking at it from a CMML perspective.

Just my 2c...

Cheers,
Silvia.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.xiph.org/pipermail/ogg-dev/attachments/20080207/c18c38cd/attachment-0001.htm

Hi,

I recognize the main name behind CMML here :)

Does the redesigning of CMML allow overlapping clips ?
This is the main reason of my current ramblings about seeking.

While karaoke was one of the initial goals behind kate, it is just
a way the format can be used with (in fact, the format itself does
not refer to karaoke at all, but styles and motions).
At the moment, it is a fairly versatile text based presentation/video
composition that can be superimposed onto a video (or not).
This includes overlapping events, and animations, which I think
CMML doesn't do (and which I didn't see it do, due to it being
metadata about other data, and thus not "changing" in itself). But
it seems to be changing goals, based on docs Conrad sent me
links to.

About requirements from Ogg itself, I don't think there is anything
really - I've just listened to Ralph's talk about seeking in Ogg, and
what I'm getting from it is "well, though, a multiple seek is
required",
so that's just the way it is. Otherwise, everything sits just fine within
Ogg.

You're quite right about CSS too. It's awfully complicated (but then
I originally thought the same about CMML including HTML bits),
and I made a simpler (and of course much less versatile) style setup.
I wonder if this is a mistake.

I didn't know you could time CSS changes however, unless you just
refer to replacing the text by another text with the same actual visible
string but css markers placed differently ?

Anyway, seeking is the only bit that's really left now, it all works fine
(from the encoder/decoder lib point of view anyway), and I didn't hit
any showstoppers from ogg, hence no extra requirements. What did
you have in mind when you referred to those issues that came up when
you designed CMML ?

As a bit of background, so you understand what I'm talking about, one
of the things I can do with kate is to have (eg):
- a (TV station ?) logo in a corner
- a scrolling transcript of what is being said
- a digital time in another corner
- a channel number in a third
- and whatever else you'd need
all at the same time on screen, optionally moving around.

It's not really showable now because you need a patched version of xine
to actually view a video with the kate bitstream rendered on top of it, but
you get the idea.
Unless the changes to CMML are geared towards allowing this kind of
things too ?

Thanks