What's your opinion on the x264 article?

[patrick aljord]

Hi,

Congrats for the release. I was wondering what your opinion was about
this article written by one of the x264 dev:

http://x264dev.multimedia.cx/?p=377

Cheers,

Pat

--
You received this message because you are subscribed to the WebM Discussion group.
To post to this group, send email to webm-d...@webmproject.org
To unsubscribe from this group, send email to
webm-users+...@webmproject.org
For more options, visit this group at
http://groups.google.com/a/webmproject.org/group/webm-users?hl=en

[LiquidMeson]

Sounds like he knows his stuff, wouldn't doubt him. The fact that he
wrote so much about it though shows he has interest/hope for it,
despite him picking out the flaws and basically stating it falls just
short of h.264.

[LiquidMeson]

If I understood correctly, here's a summary from the "summary for the
lazy": If bugs are now set in stone, this thing will need a fork. NOW.
- Carlos Solís

yep.

[Pascal Massimino]

Hi Patrick,

On Wed, May 19, 2010 at 4:43 PM, patrick aljord <pat...@gmail.com> wrote:

Hi,

Congrats for the release. I was wondering what your opinion was about
this article written by one of the x264 dev:

http://x264dev.multimedia.cx/?p=377

this article is very sound technically when it comes to h264. One wouldn't

expect less coming from Jason. I'd say the VP8 analysis lacks some

non-technical developments to enlighten some of the choices made:

ease of implementation in hardware is one of these. The article also

sometimes mixes remarks about the decoding process (which is normalized)

with the encoding part (which will certainly be refined and enhanced with

time).

On the technical vp8 part, i would mainly point to some weak points:

  * alt-ref frame is not necessarily displayed and does not exactly map

to simply an additional reference.

  * adaptive quantization *is* definitely possible in vp8 (this point was

later corrected in the article)

 * not-adaptive arithmetic coding makes multi-threading encoding

and decoding much easier, for reasonable loss in efficiency.

But overall, good reading!

-skal

[Carlos Andrés Solís]

- Carlos Solís

---------- Mensaje reenviado ----------
De: Carlos Andrés Solís <cso...@gmail.com>
Fecha: 20 de mayo de 2010 14:49
Asunto: Re: What's your opinion on the x264 article?
Para: Pascal Massimino <pascal.m...@gmail.com>


Hello people. I would like to comment on the actual status of the VP8/WebM project.

I'll begin by telling you a brief story. Around 1900, a man named Ludwig Zamenhof developed a language named Esperanto, hoping it to be the world's new unified international language. Unfortunately, he set its rules on stone before anyone could check or suggest corrections to the grammar. This caused a big uproar, to the point of causing a "fork" of the language, this time developed by a group of linguists (Zamenhof was just a doctor with little knowledge of languages). The new language, known as Ido, promised to correct the errors that Esperanto had, but after a campaign of bullying by Esperantists, Ido was left far less known than Esperanto. Today, Esperanto is far from being an international language, and still today, its supporters take Ido as a language that nobody should know about.

Why the story? Well, as Jason's article pointed out, it seems Google will set VP8/WebM's specs on stone before anyone can modify them to make them better. A really bad idea, in my opinion. Of course somebody would check the source code of VP8, why not wait at least a month to add changes to the final spec? Or if that's not possible, what about creating a "High Profile" on top of the base one, like H.264 did?

And about the WebM name, I suggest it to be changed as soon as possible. I mean, if I found about a program named WebM and was asked to categorize it, I'd place it in Internet instead of Multimedia. It also breaks the open-source tradition of naming codecs after people (Theora, Vorbis, Dirac). I'd suggest a few names, like Lumiere (from the brothers that invented some of the first cinematographers) or Gideon (a pun on Google Video). Since WebM's format is based on Matroska, I'd go as far as renaming the files as .mkv manually before having to use this name. Who in the whole world thought it could be a good name for a CODEC?

And speaking of Matroska, is it right that WebM only supports a subset of it? I would like to embed subtitles and extra audio tracks in a WebM/Matroska file, or splitting it in chapters like a DVD. Both tasks can be done in a Matroska file, can they be done in a WebM file?
- Carlos Solís

Hi,

Congrats for the release. I was wondering what your opinion was about
this article written by one of the x264 dev:

http://x264dev.multimedia.cx/?p=377

You received this message because you are subscribed to the WebM Discussion group.
To post to this group, send email to webm-d...@webmproject.org
To unsubscribe from this group, send email to
webm-users+...@webmproject.org
For more options, visit this group at
http://groups.google.com/a/webmproject.org/group/webm-users?hl=en

[Steve Lhomme]

2010/5/20 Carlos Andrés Solís <cso...@gmail.com>

And about the WebM name, I suggest it to be changed as soon as possible. I mean, if I found about a program named WebM and was asked to categorize it, I'd place it in Internet instead of Multimedia. It also breaks the open-source tradition of naming codecs after people (Theora, Vorbis, Dirac). I'd suggest a few names, like Lumiere (from the brothers that invented some of the first cinematographers) or Gideon (a pun on Google Video). Since WebM's format is based on Matroska, I'd go as far as renaming the files as .mkv manually before having to use this name. Who in the whole world thought it could be a good name for a CODEC?

 

Webm is not the name of the codec (VP8) but the format.

 

And speaking of Matroska, is it right that WebM only supports a subset of it? I would like to embed subtitles and extra audio tracks in a WebM/Matroska file, or splitting it in chapters like a DVD. Both tasks can be done in a Matroska file, can they be done in a WebM file?

No you can't put subtitles in Webm files nor chapters. If you want the Matroska features, just use Matroska and not webm. They don't have the same goals, even though it's the same technology.

I'm not sure browsers would even support multiple video or audio tracks. But such files would be ineffecient on the web (the goal/target of webm) so I suggest putting in the specs that only one video and audio track are allowed (ie meant to be played).

Steve

[Lachlan Hunt]

On 2010-05-20 14:49, Carlos Andrés Solís wrote:
> Since WebM's format is based on Matroska, I'd go as far as renaming the files as
> .mkv manually before having to use this name.

I'm not sure if you know this already or not, but I'm sure there are
people in here anyway who don't. You can't just change the file
extension and suddenly call it a Matroska file. It's still WebM
internally, even though WebM is based on Matroska.

The main difference is the DocType. If you change the DocType inside
the file from "webm" to "matroska", as long as you edit the EBML
properly, then you will make a valid Matroska file. The easiest way to
do this is with mkclean, which will reumx it for you.

$ mkclean --doctype 2 input.webm output.mkv

The reverse, however, only works properly if the mkv only contains VP8
and Vorbis.

> And speaking of Matroska, is it right that WebM only supports a subset of
> it? I would like to embed subtitles and extra audio tracks in a
> WebM/Matroska file, or splitting it in chapters like a DVD. Both tasks can
> be done in a Matroska file, can they be done in a WebM file?

Support for captions, subtitles and chapters is in the proposed <track>
element for HTML5. WebSRT is the current proposal, still under
development, for subtitles and captions, and it's likely that once this
format is agreed upon, WebM will be updated to permit its use too.

I'm not so sure about chapters though. There's currently no format
mentioned in HTML5 for chapter tracks, though my guess is that if
support for chapters lands in the future, we'll end up supporting the
same format as Matroska already does.

--
Lachlan Hunt - Opera Software
http://lachy.id.au/
http://www.opera.com/

[Silvia Pfeiffer]

2010/5/20 Lachlan Hunt <lachla...@lachy.id.au>:

Chapter tracks are also envisioned for HTML5 and indeed mentioned wrt
the current <track> proposal, see
http://www.whatwg.org/specs/web-apps/current-work/multipage/video.html#the-track-element
.

The creation of chapter tracks can be supported in exactly the same
way as captions - doesn't need a new format.

We still have to make some demos with the new WebSRT format to explain
how these things are meant to work. :-) It's indeed work in progress.

Cheers,
Silvia.

[tony.tang]

about vp8 not-adaptive arithmetic coding efficiency ,specifically for
coefficient encoding ,
H264's significant_coeff_flag coding and level coding , the
adaptibility looks weakly ,and choose level==14 as the critical value
for use EG0 coding

while VP8 use block local adaptibility (brand segement and local
complexity),and choose level==4 as critical value for directly token
encoding ,the designer should
do many statistics analysis for level distribution and choose the
threshold

In my opinion,we can't verdict h264 is efficient than vp8 in the
coefficient encoding part .

there need some test to verify

--tony.tang
On 5月20日, 下午1时13分, Pascal Massimino <pascal.massim...@gmail.com>
wrote:
> Hi Patrick,

>
> On Wed, May 19, 2010 at 4:43 PM, patrick aljord <patc...@gmail.com> wrote:
> > Hi,
>
> > Congrats for the release. I was wondering what your opinion was about
> > this article written by one of the x264 dev:
>
> >http://x264dev.multimedia.cx/?p=377
>
> this article is very sound technically when it comes to h264. One wouldn't
> expect less coming from Jason. I'd say the VP8 analysis lacks some
> non-technical developments to enlighten some of the choices made:
> ease of implementation in hardware is one of these. The article also
> sometimes mixes remarks about the decoding process (which is normalized)
> with the encoding part (which will certainly be refined and enhanced with
> time).
> On the technical vp8 part, i would mainly point to some weak points:
> * alt-ref frame is not necessarily displayed and does not exactly map
> to simply an additional reference.
> * adaptive quantization *is* definitely possible in vp8 (this point was
> later corrected in the article)
> * not-adaptive arithmetic coding makes multi-threading encoding
> and decoding much easier, for reasonable loss in efficiency.
>
> But overall, good reading!
> -skal
>
> --
> You received this message because you are subscribed to the WebM Discussion group.

> To post to this group, send email to webm-disc...@webmproject.org

> To unsubscribe from this group, send email to

> webm-users+unsubscr...@webmproject.org