Is Theora dead!?

[OpasanNekiLik]

Very simple question: Is Theora CODEC dead!?
Simple question expecting simple answer - Yes or No! There's no answer like "could be dead" or "not dead yet" - you should simply decide on answer Yes or No.

[nurbs]

AFAIK it's still under development, so no, the project itself is not dead. Hardly anyone is using it, but that's not news, and with VP8 around it's unlikely that Theora's popularity will grow.

[GodofaGap]

Was Theora ever really alive?

[OpasanNekiLik]

Originally Posted by nurbs

AFAIK it's still under development, so no, the project itself is not dead. Hardly anyone is using it, but that's not news, and with VP8 around it's unlikely that Theora's popularity will grow.

My question above, among other things, is very related to the fact that VP8 is around. But generally it seems to me that there's no more development momentum for Theora and, as you said, very few people use it really (and if use it then enthusiastically only).

[nurbs]

There was never much momentum to begin with. They took 2 years to finalize the bitstream and then another 4 years until they had their 1.0 release which was lacking in features, speed and quality. Not even 2-pass encoding was supported. Now, more than 2 years later, they implemented that and improved quality a bit, but the encoder still has bad performance, doesn't support multithreading and from what I read has rate-control problems.
Personally I don't think there will be a "modern" Theora encoder any time soon and the current VP8 implementation isn't that impressive either, but at least Google decided to spend some money on xvp8, so I guess there will be a much better VP8 encoder available next year.

[mandarinka]

Maybe... xvp8 is a vapourware (almost a myth) though, and we don't actualy know that Google is spending money on it. It might have just hired the ffvp8 author to work on libvpx, ffmpeg or other software.

[nurbs]

The x264 development newsletters still contain this:

xvp8 is under development: a best-in-class VP8 encoder built using the x264 framework.

That says nothing about current state and rate of progress, but there at least seems to be the intention to develop it.

[kurkosdr]

To be honest, Theora was not really alive from the beginning. But now with WebM/VP8, it's 100% dead. So the answer is "yes".

Theora was yet another GNU/project developed for ideological reasons that nobody really asked for. It's only purpose was to dodge patent issues and allow wikipedia and ubuntu to distribute/play video without patent worries. Othewise it's a pretty lame format. But VP8 can dodge patent issues too, and is a much better format, so Theora now is dead.

[tuqueque]

How easy is to say things without knowing or at least doing some previous research... Do some reading people, please!

This is copied from a fairly recent comment from Gregory Maxwell in Xhip's Theora mailing lists:

On Fri, May 27, 2011 at 11:22 PM, Daniel Hendrycks wrote:
> To my recollection, there hasn't been any talk about Ptalarbvorm. Is
> Ptalarbvorm development ditched? (I have noticed some commits for it every
> now and then, but nothing like the amount of commits last year)

As mentioned, the development was moved to mainline. However, it's
hasn't been super-active lately for a couple of reasons.

One of them being that the active developers have been focusing their
time lately on other projects, including some (e.g. the Opus codec)
which have externally imposed deadlines.

As of right now while the quality of the SVN code is significantly
better than 1.1 overall there are some annoying regressions that I
think we want to fix before a release— in particular our quality on
fades has gone down and our tendency to produce ringing on titles has
increased. The new faster speed level is somewhat broken (as a user
you probably wouldn't notice, but it's doing the analysis calculations
incorrectly— which is at least aesthetically bad if not a big
practical problem), and there are a lot of low hanging performance
fruit that would be nice to get in prior to a new release.

Of course, you can run the software today and report your findings.
Any such reports will help get out a new release…

Is true though that Theora is far behind VP8 in terms of quality. But in terms of encoding/decoding speed, Theora is probably the best out there!

As stated in the above quotation, there's still some work to be done, like the Temporal RDO development which will (directly and indirectly) improve very noticeably video quality, bringing it to the same quality of XviD... Which is a very widely used codec still. (if you want to read more about it, you can read the last demo from http://people.xiph.org/~xiphmont/demo/theora/demo9.html

I use Theora in some of my work and I'm pleased with its quality... Although I'm starting to create some content with VP8 and soon I'll stop using Theora entirely.

So to sume up, no, Theora is NOT dead... But IMHO, after those pending Ptalarbvorm developments, Xiph people should stop Volcano's plan and join forces on VP8 developments.

[nurbs]

Originally Posted by tuqueque

But in terms of encoding/decoding speed, Theora is probably the best out there!

Last time I checked Theora wasn't all that competitive in terms of encoding speed. x264 could easily beat it on faster settings limited to 1 thread because current Theora encoders don't support multithreading. I'm sure that with proper optimizations it could be much faster, but it just isn't there and I don't see the point of wasting more time on it.