Suggestions regarding fragmentation and extensions

[AlexT]

Right now Firefox, Opera and Chrome support
1) OGG and WebM (matroska) containers
2) VP3 (theora) and VP8 codecs
3) Vorbis and WAV audio

In my opinion to push an open HTML5 video package there should be one
container, video codec and audio codec. Regarding the container even
Monty from Xiph agrees that Matroska is a good choice "I'm biased and
I do think Ogg would have been a better technical choice for WebM, but
that doesn't mean Matroska is wrong or bad. It's what Google chose,
and as the maker of a competing container, I say it was a good
choice." so it doesn't make any sense to have OGG as a container as
well more on this later

Everyone agrees that VP8 is superior than VP3 in all aspects as long
as some patent situation doesn't arise.

WAV i.e. uncompressed audio does not make much sense on the web.

Some people might argue that choice is good but the thing is that
sticking to one choice for the codecs and container makes it easier to
implement for developers either in browsers or encoding software,
websites etc. Instead of educating the user on which codec/container
to use just make it simple since the alternatives do not offer any
significant advantages (VP3/Vorbis/OGG over VP8/Vorbis/Matroska).
Including support for the previous option (VP3/Vorbis/OGG) would only
lead to fragmentation where people would would be using different
combinations which would lead to implementation problems (container
might clash with the codec seeking problems etc etc) It might be
easier to convince Apple or Microsoft to implement one codec rather
than two codecs and containers and if someone ends up using (VP3/
Vorbis/OGG) it might not work across all browsers. In conclusion all
browsers should support WebM as a standard for <video> and discontinue
support for all other implementations while development is still in
the early phase and this also means removing H.264 from chrome.



Secondly regarding the extensions of WebM. When the mp4 container was
introduced you could not tell if it was audio only or video+audio
until apple came up with m4a and m4v which made a lot of sense. People
sometimes use different desktop applications for audio and video which
leads to file association problems. WebM could learn a lot from this
and use something like wba for audio and wbv.

Thanks for listening

[Silvia Pfeiffer]

On Fri, May 21, 2010 at 5:14 PM, AlexT <atz...@gmail.com> wrote:
> Right now Firefox, Opera and Chrome support
> 1) OGG and WebM (matroska) containers
> 2) VP3 (theora) and VP8 codecs
> 3) Vorbis and WAV audio
>
> In my opinion to push an open HTML5 video package there should be one
> container, video codec and audio codec. Regarding the container even
> Monty from Xiph agrees that Matroska is a good choice "I'm biased and
> I do think Ogg would have been a better technical choice for WebM, but
> that doesn't mean Matroska is wrong or bad. It's what Google chose,
> and as the maker of a competing container, I say it was a good
> choice." so it doesn't make any sense to have OGG as a container as
> well more on this later
>
> Everyone agrees that VP8 is superior than VP3 in all aspects as long
> as some patent situation doesn't arise.

Please note that Theora is not the same as VP3. Many improvements have
been made and it is plain wrong to call Theora "VP3" just like it
would be plain wrong to call VP8 "VP3".

> WAV i.e. uncompressed audio does not make much sense on the web.
>
> Some people might argue that choice is good but the thing is that
> sticking to one choice for the codecs and container makes it easier to
> implement for developers either in browsers or encoding software,
> websites etc. Instead of educating the user on which codec/container
> to use just make it simple since the alternatives do not offer any
> significant advantages (VP3/Vorbis/OGG over VP8/Vorbis/Matroska).
> Including support for the previous option (VP3/Vorbis/OGG) would only
> lead to fragmentation where people would would be using different
> combinations which would lead to implementation problems (container
> might clash with the codec seeking problems etc etc) It might be
> easier to convince Apple or Microsoft to implement one codec rather
> than two codecs and containers and if someone ends up using (VP3/
> Vorbis/OGG) it might not work across all browsers. In conclusion all
> browsers should support WebM as a standard for <video> and discontinue
> support for all other implementations while development is still in
> the early phase and this also means removing H.264 from chrome.

Nobody has this far managed to get Apple or Microsoft to implement
Theora/Vorbis support - it is doubtful they ever will and I don't see
a need to have to push Theora/Vorbis support while pushing for WebM
support. WebM has the chance to become the baseline code and that's
what is being worked towards, FAICT. The best you can do to drive this
forward is by using WebM everywhere and encouraging everyone else to
do that, too. Preaching to the converted is probably not going to
help. ;-)

> Secondly regarding the extensions of WebM. When the mp4 container was
> introduced you could not tell if it was audio only or video+audio
> until apple came up with m4a and m4v which made a lot of sense. People
> sometimes use different desktop applications for audio and video which
> leads to file association problems. WebM could learn a lot from this
> and use something like wba for audio and wbv.

I'd actually like to support this. If WebM is used for WebM/Vorbis
files as well as WebM/Vorbis/VP8 files, I would suggest defining
different extensions and different mime types for these. We had to do
that in Xiph for Ogg Theora files, too, because platforms use the
extension to decide which applications to use to open them and what to
do with the files generally. It simply makes life easier to provide a
file extension that minimally signifies if its video or audio only.

Regards,
Silvia.

[Chris Double]

On Fri, May 21, 2010 at 8:46 PM, Silvia Pfeiffer
<silviap...@gmail.com> wrote:
> I'd actually like to support this. If WebM is used for WebM/Vorbis
> files as well as WebM/Vorbis/VP8 files, I would suggest defining
> different extensions and different mime types for these.

A mime type for audio only is already defined:

http://www.webmproject.org/code/specs/container/

I agree about the different extension. It'd be nice to have the audio
files open in a music player than a video player for example and on
most platforms that's driven by extension.

Chris.
--
http://www.bluishcoder.co.nz

[AlexT]

On May 21, 4:46 pm, Silvia Pfeiffer <silviapfeiff...@gmail.com> wrote:

> I don't see a need to have to push Theora/Vorbis support while pushing for WebM
> support. WebM has the chance to become the baseline code and that's
> what is being worked towards, FAICT. The best you can do to drive this
> forward is by using WebM everywhere and encouraging everyone else to
> do that, too. Preaching to the converted is probably not going to
> help. ;-)

> Regards,
> Silvia.

The main point in my argument was that browsers should drop support
for the theora codec and the Ogg container and the wave codec so that
the only path is WebM. If that happens no encouragement will be needed
as WebM will be the only path available thus prevent fragmentation.
Given the short time theora support has been out I highly doubt anyone
has converted to it and even if they have they will soon realize that
WebM is far superior and the standard to follow for the future.
Regards

[Fabricio Zuardi]

IMHO browsers should keep all doors open, and I would go further and ask that browsers should also support an extra open codec, Ogg/Dirac

On the current patent uncertainty environment, having support for a single option (be it webM or theora) as far as I understand it, can be risky for the future. While on the fragmented scenario, in the future if patents make Theora or WebM unfeasible, people would still have third option to go to: Dirac.

just 2 cents

[]s
Fabricio C Zuardi
http://fabricio.org

[Alex Converse]

On Fri, May 21, 2010 at 9:45 AM, Fabricio Zuardi <fabr...@fabricio.org> wrote:
>
> IMHO browsers should keep all doors open, and I would go further and ask that browsers should also support an extra open codec, Ogg/Dirac
>

Dirac is a mess, and wavelet codecs have never lived up to their promise.

And why put it in Ogg which is basically the worst container out there
when the WebM matroska-derived container is not the issue?

Regards,
Alex Converse

[Fabricio Zuardi]

Choice and compatibility.

It should be up to the content providers the decision of which format to use, the browser is only the tool to view it. IMHO a browser that can open a wide range of file formats is better than one that can only open a small set of formats. People will perceive the one with limited support as broken and potentially switch to the one that can open all kinds of junk. Are you in favor of dropping support for Ogg on VLC as well? :)

[]s
Fabricio

[Jean-Baptiste Kempf]

Hello,

You know what? I wish we could.
But I don't have the time to fight with the upset users that will come
of this move.

--
Jean-Baptiste Kempf

[Alex Converse]

But content providers aren't clamoring to use Dirac. Should half dozen
other formats that underperform and nobody uses be added too? To
standardize on everything is to standardize on nothing. Do people
actually have plans to use Dirac for something? If so let's hear it.
Or is this simply choice for the sake of choice?

As far as Ogg the container goes, nobody seems to like working with it
except Monty. Legacy content that people don't want to remux is one
thing, but there is very little legacy Dirac content, and most of it
isn't suitable for web use.

--Alex Converse

[Fabricio Zuardi]

True, most content providers aren't clamoring to use Dirac. I use it on my website mainly as a safety measure, just in case the patent time-bomb comes to explode.

I don't know how realistic is to expect theora or vp8 to become compromised any time soon, but in the eventual case that it does happen, all I have to do is remove the access for those versions of the video clips and maintain the already generated dirac versions.

For someone that is starting a video library right now, it doesn’t hurt to distribute the eggs in several baskets, I will not have to re-generate everything in the future if/when the only true patent unencumbered format left turns out to be Dirac…

And I see what you are saying about WebM being a better container than Ogg, I am no video formats expert by any means, I mentioned Ogg/Dirac just because that's what currently available tools like ffmpeg2dirac generates today. I would love to hear more about who is doing WebM/Dirac or something better and the user agents that supports it, if any.

[]s
Fabricio

[AlexT]

On May 22, 2:23 am, Fabricio Zuardi <fabri...@fabricio.org> wrote:
> On May 21, 2010, at 12:02 PM, Alex Converse wrote:
>
>
>
> > On Fri, May 21, 2010 at 10:10 AM, Fabricio Zuardi <fabri...@fabricio.org> wrote:
> >> On May 21, 2010, at 10:58 AM, Alex Converse wrote:
>

The last time I used dirac the results were extremely disappointing.
Let's not make this a codec debate
I will reiterate what the main point of this discussion was

VP8 is better than any freely available codec out there (efficiency/
quality). Theora might have a slight advantage in low bandwidth
situations but that is not reason enough to keep theora around, the
benefits of avoiding fragmentation by far outweigh such small
advantages. Other open codecs like Dirac and Snow are no where near
good enough so it doesn't make any sense to discuss them. If you are
worried about the patent issues then you should be worried about all
codecs because according to MPEG-LA you cannot create a video codec
without infringing their patents, so that makes the patent point moot.
VP8 is the best codec we have and that should be the path forward
instead of splitting development/evangelism/etc time on theora and
others.

The OGG container is good but for whatever reasons matroska was chosen
and it even has the blessing of the lead Xiph developer.

Vorbis has proven over the years to be better than mp3 and on par with
AAC.

So we end up with VP8 + Vorbis inside matroska the best possible
situation right now.

We should just concentrate on this combination rather than splitting
time amongst other combinations which include Theora and OGG because
they will eat up some development time which would be better spent on
the main combination and would come with the need to educate users
regarding the differences between different codecs etc and would make
new implementations for new browsers/applications based around HTML5
<video> more cumbersome for the developers.

Besides choice (for an inferior codec) I would like to know what
arguments are there to keep Theora/OGG around.

Regards

[Chris Double]

On Sat, May 22, 2010 at 5:54 PM, AlexT <atz...@gmail.com> wrote:
> Besides choice (for an inferior codec) I would like to know what
> arguments are there to keep Theora/OGG around.

Who are you addressing this request to? This is a WebM list for
discussing WebM - I don't think it's worthwhile discussing other
codec's or formats support in products. Maybe discussing the future of
Theora would be better on a Theora mailing list. Or if you want
software producers not to support it contact them directly perhaps.

Chris.
--
http://www.bluishcoder.co.nz

[Valeria Zysman]

I do support Ogg as a WebM container. To have two free & open video
formats will not hurt anyone, do not you think so?. As a matter of
fact, I really hope The WebM Open Media Project staff include FLAC
support instead of WAV (or add FLAC support, anyway). ;)

AlexT said: „It might be easier to convince Apple or Microsoft to
implement one codec rather than two codecs and containers”.

This has not got anything to do with „convince Apple or Microsoft”.
They are proprietary software producers and we want a free & open
video format. So, I think we actually do not have to convince anybody
because this cause is all about good willing.

AlexT said: „The main point in my argument was that browsers should

drop support for the theora codec and the Ogg container and the wave

codec so that the only path is WebM”.

I cannot believe you are really asking for drop support of Theora &
Ogg... o_O What is the benefit of it? Theora is definitely not a
yesterday-done codec, its creation did not take a few months, but
whole years, did you know that? You've got to respect other people's
efforts, specially when they are so unique as Ogg is.

AlexT said: „Given the short time theora support has been out”...

Say what...?! What about Ubuntu, openSUSE, Fedora and many, many
others GNU/Linux distributions already using it? There are millions
and millions users out there using Ogg Theora/Vorbis (.ogv/.oga) files
right now. As you can see, this is not about „short time”, it's about
how many users are using them nowadays.

AlexT said: „I would like to know what arguments are there to keep
Theora/OGG around”.

Drop the support for Ogg Theora right now and name me at least one
useful, free video codec and one useful, free video container...

We actually don't know how much time WebM will take to be fully ready
and standardized. I know you like this project, AlexT (so do I), but
WebM should not mean an Ogg-killer because of WebM itself.

Right now (and I mean RIGHT NOW) Ogg Theora and Ogg Vorbis both are
closer to be the main, standardized, free formats than WebM. In short,
Ogg Theora is the current free, open web-video format, and it will be
at least for several months until WebM reaches mature. In the
meanwhile, we just can not wait until that moment comes. So, keep
supporting Ogg Theora & Vorbis is such a really good idea, in my
humble opinion.

Best regards,