Google to open source VP8

[kosmonaut]

According to Ryan Lawler at NewTeeVee, Google is going to announce that they're opening up the code to On2's VP8 at their developer conference I/O next month.

Crazy! Everybody's been talking about this, but I really didn't think it would happen. My questions: How good will VP8 actually prove to be in widespread application? Will MPEG-LA make good on their threats to enforce H.264 patent claims against Ogg Theora now that it could be a real contender? Can an improved Ogg Theora really do anything at this point to slow down the momentum of H.264 adoption in hardware?

Looks like the codec world just got a bit more interesting!

[Dark Shikari]

Already answered all these questions a month or so ago

[kosmonaut]

Ha! DS, you did answer the questions, but we just have to wait and see how long it takes the rest of the world to catch on.

[Midzuki]

Google to open source VP8

Great! At last, another VfW DLL to the VP* series.

[MousePin]

Originally Posted by Midzuki

Great! At last, another VfW DLL to the VP* series.

That is what I think

[invisible]

Originally Posted by Dark Shikari

Already answered all these questions a month or so ago

I know this is a little late and not exactly on topic with the blog but about moonlight: the only reason MS supported moonlight was as a distraction. It is not functional because it lacks the DRM scheme that is pretty much the whole point of it. (can't play netflix on moonlight and that's unlikely to change)

[forextrader07]

Google must add some technology to advise on code. As a developer of code knows, importing new technology is not foolproof after reading a document. It needs further ongoing advice from the maker of the technology when the developers adopting it run into a question. This is critical, and I think any company or browser wanting t adopt this tech will think or have to think about the questions that will need to be answered so there is no bugs preventing implementation, especially since bugs increase cpu and ram usage and slow the sw performance. it will all depend on how easy google makes it for developers to implement I think.

[Dark Shikari]

Originally Posted by butterfly320

apple/h.264 fanboys

What in the world does Apple have to do with H.264, besides distributing one of the world's worst H.264 encoders?

That quote is about as sensible as "IBM/Firefox fanboys" or "Microsoft/Linux fanboys".

Originally Posted by butterfly320

They indicated that there’s nothing stopping someone from hardware accelerating Theora, companies just aren’t doing it

If it's so easy, why don't they design an ASIC and publish it freely for companies to implement?

[UED77]

Dark Shikari's post was strangely prophetic on the matter. While open-sourcing VP8 is a welcome gesture, what does that mean anyway? Open-sourcing the VP8 encoder and making it downloadable free of charge does not make a free format -- it'd merely give the world the opportunity to encode video into a proprietary format whose only claims of superiority over H.264 come from On2's own marketing materials (and most press outlets that covered this story seemed to ignore that fact to up the hype).

Though I'm not familiar with all of the facets of the On2 buyout, I'm presuming all of On2's intellectual property, including patents on techniques used in the VP* family are owned/managed by Google now. However, the age-old argument against third-parties (such as browser vendors) providing out-of-the-box decoding support for VP8 still stands -- how can they be sure no submarine patents exist that one judge in the Eastern District of Texas may find to cover some method used in VP8? Bitsteam-frozen Theora has been around for 6 years now, and Apple and Nokia objected including it in the HTML5 specification, due to the same risk of submarine patents. These same companies will never implement or adopt a freshly exposed VP8, as it is even more risky in their view than Theora is.

So unless Google conducts an extensive patent search, and buys out any potential holders of related submarine patents or aggressively tries to invalidate them, the web will get another format that will only play on certain browsers (Google Chrome... perhaps Mozilla if Google keeps pulling some strings), and will likely offer a washed-out picture that is going to be a joy to see on Youtube non-HD. While normally I'd say "more choice" is good, "choices" that needlessly fragment the web video field are... questionable.

And, if by some miracle, Google does do all of the above and can 100% guarantee that VP8 does not infringe on any known and hitherto unknown patents (which is hard...), they should really work on the quality of the encoder before VP8 becomes the de facto web video standard. I'm sure On2 investors from the codec's proprietary days were impressed by PSNR graphs, but the days of blurry video being billed as "better" than slightly blocky video need to end. I was glad when x264 took a page from Xvid and decided that details are best preserved (the blocky vs. blurry grass comes to mind).

[UED77]

Apologize for the "double post", but a combination of VP8 video and Vorbis audio in a restricted Matroska container has been announced by Google, Mozilla, Opera, et al. as the "WebM" format.

Most curious is their license terms, which are BSD-inspired, and grant a "perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable" license with the exception that if you file patent litigation against them, the license is revoked. Let's hope it doesn't come to testing whether this holds up in court.

And most relevant to web video is that all new (as of May 19, 2010) YouTube videos automatically get transcoded to WebM also. Their stated intention is to transcode the entire YouTube catalog as well over time.

In either case, congrats to Google, Mozilla, Opera, Xiph, Matroska, and everyone for accomplishing this launch.

All eyes on the MPEG-LA now... and of course, Adobe, Apple, and Microsoft for their official or unofficial response.

[nurbs]

There is already a discussion about that here: http://forum.doom9.org/showthread.php?t=154669