Jobs drops hint on Google open video codec
Steve Jobs has indicated that Apple is unlikely to embrace Google's newly-open sourced VP8 video codec. Last last night, in an email to a Register reader, Jobs — or someone with access to his email account — pointed to a recent blog post in which a graphics developer says that VP8 "appears to be significantly weaker" in terms of …
This is a real email, not a fake in case anyone was wondering - I got the same reply this morning.
It is a good thing that Apple "invented" copy and paste and included it with the iPhone. I bet Steve would have responded sooner, but the battery was dead.
Well, if Anonymous Coward says so, that sets my mind at ease.
It doesn't really matter. The <video> tag isn't really going to relevant to the mass market for at least a year.
Google's adoption of VP8 plus its decision to open source it with Adobe, nVidia and AMD on board will make it "good enough" if not better - bandwidth can be the only issue - MP3 isn't the best codec around but it's "good enough". CUDA and Steam drivers are probably not that far away and they'll make encoding and decoding go like shit off a shovel. CE manufacturers will love it as it means $1 (or whatever) less per camera and less paperwork.
Just because the content's owner want h264 (cos they're to dumb to know anything else). This student's paper just says that h264 is better but it's ok to avoid a legal battle against Goog (too expensive as well as Goog is now the window of the web). So it will be ok to play h264 for free until the end of the world.
@lucmars: They wouldn't battle against Google, they'd battle against someone smaller who's using the codec.
With all the licensing nightmares out there a CS major needs at least some of the basics of IP law. In fact, often much more than the basics. C'est la vie
he says it's better than theora, for example, and it seems to be better than h264 _used_ to be
it's the patent-encumbrance that's the real killer though.. there should be a law like with trademarks where if you see someone violating your patent, you can't just hide until a bigger target comes along, you have to take action
if you hold a patent, you either defend against all violations or none, you can't pick and choose who and when to defend against. just as "prior art" sets a precedence, so do your actions if your defending: if you don't defend that sets a precedence too.
if you don't react to a violation within xx time after knowing, you forfeit the right to react at all.
"if you don't react to a violation within xx time after knowing, you forfeit the right to react at all."
...for very large values of xx.
Clearly Google wouldn't have splashed out $125m for something that's crap.
I trust Google to have spent wisely (as they have always shown in the past), over the opinion of someone with a heavily vested interest in H.264.
Jobs must be desperate to use that as his cite... Seems the Apple viral spin machine is in overdrive. Can't blame them, as even Microsoft have backtracked on their no VP8 stance now, and stated than IE9 will support it via a plugin. (the usual Microsoft 2 small backtrack steps are less noticable than one big about turn), so it's just Apple left.
I can see iPhone 4 now... No flash, no HTML5 Video either.... (well it will support H,264 but nobody will be encoding for it).
yes, we all trusted Google to use their StreetView cars without trespassing and violating people's privacy or snooping all their private emails from their wifi - but this snug warm feeling of trust didn't actually stop them from doing all that did it?
just because Google are so big and mighty and have the monopoly over the web doesn't mean they're incapable of making a mistake
@jai
You seem to have missed the rest of the relevant information in the Google WiFi story. They only captured data on OPEN WiFi networks. They didn't run round the EU cracking WEP keys or brute force WPA keys. Sure they did capture data without permission but likely your neighbor is already surfing porn on your open WiFi anyways.
Last I checked there are more search engines out there than JUST Google. I can think of at least two others which hold a decent market share percentage. You may know them Yahoo! & Bing. Now I am not saying they are as good as Google at search but that being said you were not arguing competency.
I read his x264 blog post - he just comes across as someone who is very p1ssed with VP8, and misses the point by a country mile. VP8 is not meant to be better than H264 (which will still be the codec of choice for offline movie encodes etc.), it just needs to be good enough for free web use and from what I've seen of VP8 it's plenty good enough for that.
exactly, it's as if people think that there can be only one codec for EVERYTHING in the entire world of video: this is about WEB video, and that means STREAMING-a live bandwidth issue.
for those that want to have their films and TV series to watch, etc no one is saying yu can't use better codecs. use the best options you want, it's your content in your house on your hardware, etc
but this battle is for the web browser and the internet content that it will have to decode: will it be with plug-ins, what codec is used, what language will the W3 standardise, etc
as you said, for online web content, VP8 is plenty good enough. if web content wants better quality it can simply use more bitrate, but the majority of it is lower quality and therefore "good enough".
jeez, so it's *only* good enough for ~95% of the web, is that last 5% *really* worth giving up legal and creative freedom ?
That's how it read to me.
I'm very interested in the technical differences, but the venom with which he approaches VP8 is hilarious. It's an extra, free, codec. What's wrong with that?
It seems to be at least as good as h264 in most situations according to even his analysis, and to me, that sounds good enough.
The guy specifically SAYS that it's probably good enough, if you actually read his blog post...
he's not so bad, once you get to know him.
^_^
Once again good ole Stevie boy cites another reputable source as to why something is bad or good. The last time it was some hack blogger. Now he cites a 3rd year college student whose opinion should obviously be taken over the geniuses at Google that did what I'm sure was an extreme amount of research before committing $125 million for VP8. Keep up the good work Steve so that eventually the intelligent masses will simply start to ignore you.
I think he's given up the marketing these days as well.
VP8 is worse than H.264 videos encoded with the "high profile" mode...but it is worse than the baseline profile, which is used for youtube and it's the mode implemented in iphone. So youtube video quality is going to _improve_ with the change....
Jason Garrett-Glaser is a really good dev, but as a x264 developper the blog post *have to* be biased.
http://lists.wikimedia.org/pipermail/wikitech-l/2010-May/047795.html
jobs (or whoever is writing his mail) knows it and by using it is just spreading FUD. But that is generally the case with jobs mails to the crowd :)
The idea of "good enough," as stated in other comments, has entirely passed this x264 developer by. Of course, even I would bias my opinion toward something that I am actively contributing to. That said, I would agree that there are many things in VP8 that need some (or A LOT) of work, but that is what FOSS is all about: let the community do it. However, in the mean-time, the codec is "good enough." No one designs websites to be elegantly displayed on high-res 1920x1080+ LCD monitors. Likewise, no one would use VP8 to encode Blu-ray discs. VP8 is being pitched for web-hosted videos, not offline content. Therefore, it is EXPECTED to be lower-quality, grainy, blurry, etc. There's plenty of people that have no problem watching youtube videos that look like a VCR tape being eaten alive. A quite-decent picture (albeit not as sharp as high-quality h264) will most likely be more than adequate for the rescaled, mob-phone-shot vids of loser garage bands banging on trash cans in their underwear or an advert of someone brushing their teeth (it might even blur all those "white" teeth enough to make the ad believable! Bonus!).
Google are behind it, MS are supporting it, Adobe are supporting it, Mozilla are supporting it. Nvidia, AMD/ATI and ARM are to accelerate it and Intel will follow - who gives a gnat's what Steve Jobs thinks, it doesn't matter remotely .
"Google's adoption of VP8 plus its decision to open source it " - ano.
I suspect when Google starts encoding youtube using this codec, it will not matter a damm if MS or Mac don't install VP8 as a default, because almost every developer will know everybody and there dog will have VP8 installed, where as this will not be true of h264 i suspect.
I also note that VP8 might not be the end, google might put the $$ in to develop VP9 or some other sweet codex.
Not that I am a massive fan of google, but roll on with this :)
Google controls YouTube, and thus controls a great majority of the videos being streamed online. Whatever Google chooses to encode their videos in, browser manufacturers are pretty much going to have to support it, or the average Joe will be falling all over themselves, calling the IT kid in the family to remedy the situation.
W3C has a history as one of the staunchest defenders of thorough, well-conceived and well-developed standards free from contamination by self-interested outside parties. As such, it's a bit surprising that they would let pressure from the likes of Apple deflect them from their stated intention to specify a particular codec for the official HTML5 video tag, chosen, in part, because it's not proprietary. It might be another story if concerns about Ogg/Theora's performance led to a decision to hold off on the spec while they kept looking for a better solution, but I'm sure they must have realized that the decision to omit the codec from the spec altogether would leave the door open to abuses such as the standoff that Google appears to have finally broken, particularly given the long and well-established history of the two 800-pound gorillas in the scene of using their muscle for just this kind of advantage.
If they aren't doing so already, agencies with the responsibility to issue specifications like this one need to dedicate some amount of time and effort to look for flaws like this one that could be exploited by certain parties for lock-in or some other advantage, and make that very clear to any and all who might be tempted to twist things their way that they will do so.
Ohhhh, Google upset Steve's revenues from MPEG-LA's licensing of X.264. Cry me a river Steve.
As for patents, who knows? VP8 has a longer heritage than X.264 so patent litigation has a fair bit of risk attached.
It's ridiculous that the VP8 quality argument is allowed to stand when it's as good as the baseline H.264 mode, which is what everyone will have to use for web videos if they want them to work on mobile devices. The other H.264 modes may as well be a different codec which nobody can/will use.
As for Jobs: http://www.pretentiousname.com/temp/joined_up_thinking.png
Remember that Theora was based on On2 VP3 codec, and uses behind the state-of-the-art technology, but it now performs much better than the original (some claim almost as well as H.264). Similarly, one can expect that (barring devastating software patent attacks), the VP8 implementation will be improved by some very clever coders so that it will eventually be on par with H.264 for practical purposes.
Stock replies to e-mails are one thing, but this is a bit terse. No hi, no bye, no "thanks for mailing." In a bad mood, Steve? Did Google ruin your week?
Besides, it's kind of ridiculous to take an x264 developer's opinion of a rival codec and treat it as a balanced enough opinion on which to position your company in the market. Developer of established codec thinks it is better than rival codec that just appeared yesterday. In other news, bears shit in woods.
We all know what Steve is *really* thinking, and it is down to two things: control of content, and money.
As a side note, I am sat across from Mr Bloe here in Leeds, and he is grinning like the Cheshire Cat at seeing his name splashed across the tech sheets. You've made his day, El Reg. ;-)
you must be kidding - this guy isn't 'nobody' - he's one of the main devs on x264 and was working also on ffmpeg and is video encoding expert who knows what he's talking about
Obviously he doesn't get the point. Online vs. Offline.
Recently, a comparison of Ogg v H.264 was carried out and used to refute any charges that Ogg was inferior. As Googles proposed codec is meant to be superior to Ogg, I don't see how it can be significantly worse than H.264
Whether you agree or not, I will say it is good to see Steve Jobs willing to defend his POV
he's replied linking to an analysis of VP8. I don't think Apple really cares about this as history shows they've been quite flexible when it comes to implementation of different codecs.
Ogg is inferior to h.264 - my own tests and you can find comparisons on the net.
Regarding VP8 vs h264 and others, analysis has links to comparisons, from those screens it looks like VP8 is more on the same level like XviD rather than h264.
I am for one standard but I'm not sure I agree with this VP8 thingy - could face patent problems, badly written spec, quite a few bits are completely undocumented, performance wise - not a clear winner and it's just another codec.
H264 makes sense as it's been widely adopted on various devices/medias (broadcast, blu-ray, net, camcorders, portable devices..)
Yes, it's open standard but not royalty free but for most of the users it doesn't really matter, licensing applies on 100k + and is 0.02c.
"Whether you agree or not, I will say it is good to see Steve Jobs willing to defend his POV"
Or rather, link to other people defending his POV*
*(Which, I might argue, isn't his actual POV)
Steve Jobs doesn't have a PoV, he has monetary interests in mind. The day he does actually have a point of view about something like this is the day humans can live 10,000 years.
Read what you speak of and then consider arms (as in guns, not "hi mum") trade. Do you honestly think the execs of all those companies want a war in their front garden?
"H264 makes sense as it's been widely adopted on various devices/medias (broadcast, blu-ray, net, camcorders, portable devices..)"
It makes feck all sense for the free online web. Just like the 3rd year grad student, you're missing the point.
"Yes, it's open standard but not royalty free but for most of the users it doesn't really matter, licensing applies on 100k + and is 0.02c."
Yes, for now.
I don't think you are understanding what this "3rd year grad student" actually does - he is one of the primary developers on the open source x264 encoder. Have you tried actually reading some of the code that he, and some of the others in the project produces? They seem quite capable of producing software that is very fast, and still produces excellent quality of video, and still complies with the H.264 specification. I would rather take his analysis of the VP8 over your analysis.
What he basically does is compare each step of both H.264 and the VP8 codec, and comes to the conclusion that the 2 codecs share a lot of ideas. That they share a lot of ideas should come as no surprise, as the development of codecs such as H.264 happens in open forums (not open as in wiki-style, but open in that everybody can, for a fee, get access to complete specification, both during development and afterwards), whereas VP8 and similar codecs are developed completely inside the company, with no external input nor no external review. The "3rd year grad student" comes to the conclusion that because the H.264 codec and the VP8 codec share so many ideas, although not on everything, that VP8 will now become susceptible to some of the same patents that H.264 is susceptible to.
I do not want to sound like I am defending H.264 here, but there was a very similar situation not so long ago, with what became the VC-1 codec. It started its life as the Microsoft Windows Media Video 9, and was available for NDA-licensing from Microsoft. However the only company to receive money from this licensing was Microsoft, as the specification was not publicly available. They later changed their policies with regards to the codec, and published the specification, claiming that it would be free for everybody to use. What happened? Now that the specification was publicly available, everybody that had patents within the field of video compression began looking through the specification, and the result today is that the VC-1 codec is priced exactly like H.264, although Microsoft claimed it would be free for everybody.
Just because Google says that VP8 is open source, instead of being a proprietary, does not mean that it does not violate any patents. It just means that the patents holders previously were not able to review the specification to see if any patents were being infringed.
I am not defending software patents here, as I live in Denmark, and we generally have a reasonably sane patent system, but in the US, this could very well become a problem for VP8.
That is my favorite line from the article :-P
You'll note that despite the assembly being written by "retarded monkeys", he later says that "the current one seems to be reasonably well-optimized and has SIMD assembly code for almost all major DSP functions, so I doubt it will get that much faster."
So either these retarded monkeys are seriously gifted SIMD assembly programmers or the codec is going to get much faster when real programmers get a shot at it.
You can't have it both ways, but he's trying to.
You think that's scathing? Here's what he had to say in a recent codec shootout about the Apple H.264 encoder that they ship with Quicktime and use in iTunes:
"Of course, a lot of things weren’t surprising either: Apple and Badaboom have atrociously bad H.264 implementations. We pretty much already knew that."
"Above all, this graph shows the enormous importance of good implementations: a bad H.264 implementation (Apple) can lose to a good implementation (ffmpeg) of a last-gen format (MPEG-4 ASP)."
And it's not just Apple's H.264 he found substandard, he included an older codec called SVQ in this test, but not the Apple version: "This was chosen as an example of the best of the pre-transform generation of video formats, the vector quantization codecs. ffmpeg was used because the Apple version is atrocious"
Finally, if you go to results graph found at the URL below, you'll find a previously obscure codec called VP7, the predecessor to Google's newly released VP8, which soundly outperforms Apple's H.264 and keeps up with the best of the H.264 implementations.
http://x264dev.multimedia.cx/wp-content/uploads/2009/08/quality_chart1.png
All quotes from here:
http://x264dev.multimedia.cx/?p=102
The article Steve Jobs referenced (from an H.264 developer -- perhaps not the most impartial judge of things?) is fallacious as it uses "straw man" arguments to try and convince people that the patent-encumbered H.264 codec is "better" than the free, open-source VP8 codec from the WebM project.
- The H.264 developer is comparing the performance of a beta release of the VP8 code to the current H.264 code.
However, the WebM project page specifically states that “there is more work to be done...some features of the WebM format are not yet complete...we expect to achieve better visual quality and performance in an official release soon...the performance of VP8 is very good in software, and we’re working closely with many video card and silicon vendors to add VP8 hardware acceleration to their chips”.
- The H.264 developer states that the VP8 specification is *final*.
However, the WebM project page specifically states that “the code and tools can evolve and improve for many years without requiring changes to the core specifications. We’ll maintain a separate branch of the code, however, for bold new ideas that could alter the specifications. If there are significant improvements to warrant a new revision we might adopt them, but only after careful consideration and after discussing suggested changes with the WebM community.” Open source code really never is “final”, and group consensus can agree to improve things while retaining as much backward-compatibility as possible. Google asks for code improvement suggestions on the WebM project page.
- The H.264 developer avoids mentioning the royalty penalties associated with using the H.264 codec, and that all these royalties are subject to change in 2015. And there is nothing “free” about H.264 -- someone somewhere is always paying royalties to MPEG-LA. Thus, MPEG-LA is effectively asking the world's citizens to slip the H.264 noose around their necks by allowing H.264 to become a web standard, with absolutely no guarantees that H.264 royalties will not skyrocket in 2015, or be preferentially enforced to effectively exclude certain people or agencies from being able to exchange information using that format.
In any case, MPEG-LA is obviously worried, as WebM including VP8 – combined with the recent announcement of a Digital Agenda by the EU for open, interoperable digital standards -- has the power to completely wipe proprietary, patent-encumbered file formats off the world-wide web. And that would be a good thing IMHO,for the sake of free and open communication between all the world's peoples. I think that Apple, Microsoft, and other MPEG-LA special-interest groups may continue their half-hearted FUD attacks against WebM, but would think more than twice about attempting a patent attack (and thus remove themselves from the ability to support and use a codec that will most likely be supported by the EU and the rest of the world, according to the clever license chosen for VP8).
Apple has a history of attacking competitors (and even ex-business partners like Adobe) with any tools at hand, so allowing a time bomb like H.264 to be adopted as a "rich peoples' club" web video standard would probably not be in the world's best interest, again in IMHO.
There is a simple way around this problem for MPEG-LA: just open-source H.264 with a BSD-style license similar to what Google used for VP8. Get rid of the royalties, and provide it free to the world. But the money-grubbing patent holders and MPEG-LA seem to be too greedy and too power-hungry to follow Google's example and make a gift of free speech and free communication to the world. So I for one will look forward to a WebM-centric web.
http://www.webmproject.org/about/faq/
Of course if a compression method is better, it will run slower! Duh!
Did he have to study for 3 years to come up with that?
The thing that google is planning on, is the constant increase of our CPU power..
So, soon what appears sluggish today will be lightning fast -- except if you still use Windows, of coıurse :)
-turgut
İzmir İnstitute of Technology
Turkey
I am truly amazed that one of the richest men in the world would have to rely on the opinion of a third year comp student to reinforce his opinion. I mean...a third year comp student. Really?
Should he ever again need a liver transplant I think he should get the opinion of a third year medical student.
But then, if the e-mail came from Mr. Jobs, he knows whats best for fanbois. He makes the right decisions for fanbois, and when he speaks all fanbois fall in to line. Since he has fanbois trained like seals, he naturally commands and expects the creators and holders of technology to do as they are told.
Mr. Jobs certainly has one amazing ego, and you'd think that his brush with death would have humbled him down to the same level as all of us human beings.
Some of you doesn't seem to notice that MPEG-4 AVC (aka h.264) is a _standard_ which happens to have a reference implementation, which ISN'T x264. x264, instead, is an _implementation_ for the video encoding process compatible with the h.264 standard.
And AFAIK, x264 doesn't pay any royalties nor licences the h264 spec from MPEG-LA, despite their willing to do so in the future and sell their licences to anyone willing to use their encoder commercially, as the h264 MPEG-LA licence states the free (as in gratis) use is non-commercial only.
Also, some of you doesn't notice that like x264 (the encoder) and many h264 decoders (e.g. ffmpeg's) are in fact open-source and free, much as VP8 is. But they aren't patent-free, as they publicly encode/decode bitstreams compatible with the h264 standard. On the other hand, VP8 claims to be it is patent-free, but it couldn't be because no one made a patent trial yet.
Regarding VP8, I think Google would do better if they paid developers to work on Dirac standard/Schroedinger codec and develop a next-gen codec/standard.
"This probably can’t be improved that much," he says, referring to VP8 decoding capabilities. Google is unlikely to change the VP8 spec, but it and partners will continue to develop the software itself.
So H.264 has some implementations that are far better than the others (according to the article) yet VP8 probably can't be improved that much? Just because he cannot see how doesn't mean it cannot, especially given the variance in H.264 implementations. I'm sure when others get their hands on it improvements will come. Let's not forget that H.264 is a pig on lower spec'ed hardware and needs GPU offloading - atoms struggle/fail with 1080p and Ions don't.
I can't see Google having pissed $125m up the wall for nothing.
