H.264 Licencing and Free Software
weblogs.mozillazine.org
weblogs.mozillazine.org
I understand that there are possibly technical reasons for Chrome\Safari choosing H.264 as the default codec instead of Theora but it strikes me as somewhat disingenuous for people to harp on about the need to move away from a proprietary technology like Flash, if the alternative is an open standard built on top of a proprietary codec.
Anyone can view the standard and implement it. To distribute, you are to acquire a license from the MPEG LA. The terms are not onerous or the fee excessive. Google and Apple both pay the license fee for each download of Safari or Chrome. The patent holders have so far not prosecuted projects like x264, which distributes its implementations without a license.
Do I think software patents should be abolished? Yes. Do I think Mozilla is holding back the open web by not supporting H.264? Yes.
"Apple Computer has rejected proposed licensing terms for the emerging MPEG-4 video standard, leaving the future of its QuickTime multimedia technology in limbo.
At a company-sponsored conference on Tuesday, Apple previewed the latest versions of its QuickTime digital media player and new QuickTime broadcast technology, both built on MPEG-4. But the company said those products will remain grounded indefinitely as it seeks to change MPEG-4 licensing terms unveiled last month by a key licensing clearinghouse.
The terms, set by MPEG LA on 31 January, have prompted a wave of criticism in streaming circles for including a per-minute streaming charge that adds up to 2 cents an hour for video clips delivered in the MPEG-4 format."
from Apple's MPEG-4 snub may halt QuickTime
http://news.zdnet.co.uk/internet/0,1000000097,2104282,00.htm
If Apple can fight against stupid licence terms why can't Mozilla?
I've also heard that the only reason Microsoft's VC-1 codec is in the Blu-Ray standard is because MPEG-LA were playing hardball on their H.264 licence fees.
Here's the manager of Blu-Ray making the very simple point that surrendering to a single patented standard is only going to let them ratchet up prices:
"There are other advantages to having two compression techniques," Doherty says. "Moving forward, the technologies can have competition to help drive down licensing fees and improve picture quality."
from here:
http://www.pcworld.com/article/117681/bluray_disc_to_support...
If Apple (who also fell out with them about AAC payment terms) and Blu-ray need to play games to achieve reasonable terms then what hope do the small businesses have, especially when the standard response is "let's all lock ourselves into this one codec, what could possibly go wrong?"
If we're lucky, this is all actually maneuvering on the part of Mozilla to get the MPEG LA to lighten their terms with Mozilla.
Apple and Google have already tried to get MPEG-LA to freely licence a lower quality subset of H.264 to break this impasse. The individual patent holders of that subset weren't impressed by this suggested compromise and even if they were it would still be a bait and switch technique that leaves open source products as second class citizens.
One more thing, you call it an open standard elsewhere. If a standard is described as "open" then that traditionally means no patent fees. H.264 is a standard, but it is not generally held to be an "open standard": http://en.wikipedia.org/wiki/Open_standard
>Do I think Mozilla is holding back the open web by not supporting H.264? Yes.
I'm wondering if it's fair to expect Mozilla (which I'm given to understand is a non-profit organization) to pay a license fee on every download the way a commercial entity like Google or Apple does? Do they get enough donations\royalties for this to be feasible?
Edit: Mozilla's explanation: http://www.osnews.com/story/22787/Mozilla_Explains_Why_it_Do...
Unless their primary fear is of being a larger patent attack surface.
Additionally, I was under the impression that the patent sabre-rattling can happen only in places where software patents are valid. So why can't Mozilla offer a build of Firefox for download only in, say, India? Too much effort?
"... Hence there is not need to obtain any patent licenses for VideoLAN software within the European Union. In other venues, please check any applicable law." http://www.videolan.org/support/faq.html
The VideoLan organisation may be based in the EU, but there are a number of companies in the US whose servers act as VLC mirrors. Would they be potential targets if the MPEG-LA decided to do something about VLC being unlicensed? With its widespread popularity it can hardly have gone unnoticed.
It's going to be GIF all over again, oh joy.
There is another advantage VLC has: it is academic project, sponsored by University. Even in US, you can use patented technologies for research without license.
In your last sentence you are right, you can but you can't pretend you are aiding others to do their own research as a cover for commercial activity as that would be contributory infringement.
Then the browsers that don't support Theora are the ones that only have a subset of support.
Embrace, extend, extinguish and all that.
This is entirely incorrect. Despite the wording of the EPC disallowing patenting of software "as such". That little couple of words on the end means that you can patent software - the old term was that it required a "technical contribution" or "technical effect" though there have been decisions on this since I was working in the field (particularly Aerotel/Macrosan in the UK).
http://www.ipo.gov.uk/p-policy-computer.htm is a reasonable and short summary.
Suppose my neighbor exceeds the maximum driving speed regularly but so far nobody has come after him. I am the owner of a truck company. Shall I tell all my drivers, "yeah go ahead and exceed the max driving speed, they're not coming after you"?
No matter how low the fee, it will rule out most open source and hobby projects.
The hobby projects which use H.264 have not been prosecuted, as far as I know, though I feel it's wrong for this possibility to exist.
You don't have to compile chromium any differently, just tell it not to use its bundled ffmpeg (even chrome only includes a few codecs), and it'll figure out what's supported at runtime.
Nobody is going to come after you. You're not even worth their time. Is it stupid that technically this may be interpreted as a possibility under law? Yes, but that's the stupid country we live in. The same thing could happen with Theora if someone claimed to hold a patent covering it, and you'd be equally screwed.
And what if I put the source code of this application on github or any other repository that's open to anyone?
No problems there.
And if I want to found a company that wants to make a living working with video online but nothing charging users directly, will I have to pay royalties only because every new user will expect that I should support the patented codec that he uses?
Talk to your lawyer. You're a company, right?
Now, what's about that's is open?
The standard is open. You are free to implement it. The licensing terms are open. You are free to license it. You are not free to distribute implementations to users without licensing the technology that has been openly provided.
This is key. Theora is just as likely to be a patent target as h.264. You really can't do a competitive video codec right now without stepping on somebody's patent.
Ignorance is a defence here, particularly for end-users, as this is only potentially tortuous infringement and not criminal. Even for a major corp, if they can't be shown to have been aware of the patent then they can avoid a far greater penalty.
You can be told to cease and desist on the basis of a submarine patent on Theora, for sure, but a compliant end-user losing money over this? Not a chance.
Incidentally patents only cover commercial activity (at least in Europe).
Yes it is. It is owned by a consortium, and that consortium wants money off me if I write and distribute software (my own software, that I have written, for fucks sake, not their software that they have written) that implements it.
> The terms are not onerous
The license terms are incompatible with free software. That makes them onerous.
Proprietary means "manufactured exclusively by the owner of intellectual property rights (IPR), as with a patent or trade secret." So, no, it's not proprietary. The fact you can write your own version of the standard without reverse engineering proves that.
H.264 wasn't trivial to create. It involved a lot of work to both meet the quality requirements and the technical requirements for supporting all the various ways in which it can be used. MPEG-LA isn't imposing a fee for your work, they're only imposing it for the costs associated with defining the standard. You're not charged more if you write a more complex implementation of the standard.
That being said, I still think it's dumb to charge for the standard. But it's perfectly within their reason and we have OSS alternatives to fall back on.
In what sense is this not proprietary?
http://dictionary.reference.com/browse/proprietary
It seems to me that MPEG LA is the proprietor, and they are controlling access to their property.
correction: It's just Apple/Safari that does not support Theora, not Google/Chrome, which does.
The reason H.264 is more secure for Apple and Google is that the technology is already covered by an existing patent, which is licensed out by the holder. If someone wished to claim ownership over technologies covered by H.264, they would have to go up against MPEG/MPEG LA and many others.
With Theora, as far as I know, the burden for legal defense would fall on whoever is using it, namely Apple and Google. The situation is ridiculous, but there you go.
One of the avenues Apple was exploring in order to discount Theora was checking whether the Theora patent licence applied to independent implementations or just the xiph version.
Google made that point moot by buying the original patent owner.
Not true. They've chosen to implement it in a codec-agnostic way. It simply requires installing XiphQT, which itself only involves moving a file into a folder. This is unlike (say) Mozilla, to which <video> support of h.264 cannot be added by any means.
You could argue that Apple have made it the solitary option on their mobile devices, but that's because it depends on the availability of hardware decoder support, which doesn't exist for Theora.
and the argument would be moot.
No, the argument would shift over to trying to get content creators to use Theora, even though 70% or so of the browsing audience doesn't support it or <video>. Web video would still be dominated by h.264 in Flash, the only format that would work everywhere.
iPad, iPhone and some other mobile devices cannot run Flash content.
The Readability bookmarklet works well on this.
Asa Dotzler's blog (http://weblogs.mozillazine.org/asa/) has the headings twist by five degrees or so when you mouse over them; he's not even a Gecko developer, so he's got no excuse.
.title a:hover {
-moz-transform: rotate(-3deg); -moz-transform-origin: top middle;"
color:#2A0000;
}
If the OS vendor already pays license fees for the decoders, than that should cover it. If the vendor doesn't, well, then that's a problem.
if you want to bundle a h.264 decoder with your software, you need to pay, too. Only if you use the OS' native decoder, you don't need to pay. But Firefox is multiplatform. WinXP, Linux do not support h.264 out of the box.
Clearly, the likes of Youtube and Vimeo don't care about this. They do seem to care about using a codec that is standard, widely supported and with good and varied implementations (both encoding and decoding, with or without hardware acceleration). Theora has neither of these characteristics.
"Only if you use the OS' native decoder, you don't need to pay. But Firefox is multiplatform. WinXP, Linux do not support h.264 out of the box."
Well, WinXP and Linux do not support Flash out of the box, but that doesn't seem to pose any problems.
I don't want to sound offensive but just because some tiny portion of people would prefer a free/open solution, the large majority of the people (who actually use a commrecial OS and and don't really care about the whole thing just want YouTube working on their desktop) shouldn't be "punished".
What about turning it around? 100% has access to open decoder, and if 2% wants to use licensed codec, they can. Remember, Theora was suggested as baseline codec, not as only codec.
Firefox is another story; they aren't charging, and according to the article, would be responsible for the licensing costs for every one of their (hundreds of millions of?) users.
Frankly if Safari and IE were to suddenly support Theora (or Firefox and IE to suddently support h.264), this argument would be over. No one cares about the technical details. Any codec that works pervasively will win.
They haven't done so yet because there's still room for improvements to be wrung out of the current spec and the longer they wait to do a non-backwards compatible spec break the better they'll be able to make it.
If Google releases one of the VP codecs then it's probably not going to be that much better than what xiph could have done if they weren't trying to keep compatability since they're based on the same core technology.
In general, I've been disappointed with theora output for very high def stuff.
What's more, the video standard affects more than just what software component you load. Vendors like Apple have devices which must support video playback in hardware. Without a very good solution there, theora is basically a non-starter. And people who complain that Apple should support Theora because it is free need to consider that, moving forward, mobile playback is extremely important.
http://www.schleef.org/blog/2009/11/11/theora-on-ti-c64x-dsp...
Generally, this "hardware" decoding argument is bunk. It's just software in conjunction with a DSP. If your vendor gave you proprietary software that you can't update (and maybe they can't either if it's 3rd party) then that sucks, but it doesn't magically make it "hardware".
Note the big thing holding it back from even better performance is private APIs that they can't yet(!) use in open source code, not any lack of "hardware".