Cisco going to pay H264 licensing fees to provide free codec

October 30, 2013 9:05 PM

CIsco has decided to pay MPEG LA licensing fees for EVERYONE! Cisco has decided to pay the MPEG LA licensing cap fee and will be open sourcing, as well as providing free binary blobs to browser providers, for H.264 in order to push it's adoption in WebRTC.

This is obviously a push for their video conferencing products, and I can't say I am entirely happy about the binary situation (only currently planned for X86 and ARM), but I figure something is better than nothing, right?

posted by Samizdata (43 comments total)

From TFA:

We plan to open-source our H.264 codec, and to provide it as a binary module that can be downloaded for free from the Internet. Cisco will not pass on our MPEG LA licensing costs for this module, and based on the current licensing environment, this will effectively make H.264 free for use in WebRTC.

Reading that carefully, I'm not entirely sure if Cisco will cover the license cost for anything except use of their binary module in WebRTC (e.g. compiling their source and using the codec for something else isn't a use within the provisions they've paid for). No?
posted by snuffleupagus at 9:20 PM on October 30, 2013

Is there a reason why they can't pay the licensing fees for code that I compile myself?

Also, does this mean that Cisco (along with Google) are now officially on record as stating that codec patents are stupid and bad for business? We can't put nails in that coffin quickly enough...
posted by schmod at 9:21 PM on October 30, 2013 [2 favorites]

I agree with that. I can has G729?
posted by snuffleupagus at 9:22 PM on October 30, 2013

It puts H.264 into Firefox. The <video> tag now has a universal format and that is H.264.

although I'm sure Wikipedia will continue their WebM crusade.
posted by Talez at 9:22 PM on October 30, 2013 [1 favorite]

Too bad they didn't go straight to h.265, but I see why h.264 is easier right now.
posted by Joakim Ziegler at 9:27 PM on October 30, 2013 [3 favorites]

I agree with that. I can has G729?

Opus will take over the world soon anyway.
posted by Talez at 9:30 PM on October 30, 2013

Comments from Brendan Eich, CTO of Mozilla: Video Interoperability on the Web Gets a Boost From Cisco’s H.264 Codec
posted by gen at 9:40 PM on October 30, 2013 [1 favorite]

Comments from Monty Montgomery of Xiph.org: Comments on Cisco, Mozilla, and H.264
posted by gen at 9:41 PM on October 30, 2013 [2 favorites]

Why is it that every video codec, even supposedly open-source and free (beer & speech) has to worry about MPEG LA? Did MPEG LA apply for overreaching patent(s) that cover all digital reproduction of moving pictures with sound?
posted by slater at 9:41 PM on October 30, 2013

Related: Open codec pioneer leaves Red Hat, joins Mozilla to work on next-generation video codec
posted by gen at 9:42 PM on October 30, 2013 [1 favorite]

Why is it that every video codec, even supposedly open-source and free (beer & speech) has to worry about MPEG LA?

See Forbes (no less): Patent Pools: Do They Kill Innovation?
posted by gen at 9:44 PM on October 30, 2013 [2 favorites]

slater: yes. MPEG LA and other technology groups have patent portfolios that are wide-ranging and obfuscated.

Part of my job deals extensively with providing test tools for MPEG LA and other patent portfolios. It makes me feel a bit icky, but I figure that when the most innovative work in my industry happens in patent licensing battles, it's a sign that the entire industry is on its way out.
posted by infinitewindow at 9:49 PM on October 30, 2013 [1 favorite]

Thanks for the info, gen & infinitewindow!
posted by slater at 9:51 PM on October 30, 2013

Joakim Ziegler: " but I see why h.264 is easier right now."

Not really. WebRTC support is so spotty at the moment, that if we somehow got the few key players to agree "WebRTC will use Theora/WebM/h265, and that's that," it wouldn't really affect all that much (apart from stranding users of devices that rely on hardware-based H.264 decoding). There's not really any meaningful installed base to break compatibility with yet.

That being said, the features of h265 probably won't be terribly useful for videoconferencing. (Hell, h264 has a lot of nifty features that aren't widely-supported, but greatly improve quality or compression ratios. It's *still* generally not safe to use CABAC if you want to widely distribute a file among a wide range of devices.)
posted by schmod at 9:51 PM on October 30, 2013

Opus will take over the world soon anyway.
...
slater: yes. MPEG LA and other technology groups have patent portfolios that are wide-ranging and obfuscated.


Digium's developer note on Opus' non-inclusion in Asterisk is interesting in this context.
posted by snuffleupagus at 9:53 PM on October 30, 2013

(apart from stranding users of devices that rely on hardware-based H.264 decoding)

Which, from a practical perspective, is almost every pre-Tegra 4 smartphone.
posted by Talez at 10:05 PM on October 30, 2013

Why is it that every video codec, even supposedly open-source and free (beer & speech) has to worry about MPEG LA? Did MPEG LA apply for overreaching patent(s) that cover all digital reproduction of moving pictures with sound?

MPEG LA isn't a single monolithic entity. It's made up of dozens of companies performing research on multiple ways of doing video compression. And when they figure out shit that improves the status quo they patent the hell out of it as a matter of course. So when some Fraunhofer-Gesellschaft engineer figures out some esoteric way of doing quantization or something, even if it doesn't quite make it into the baseline standard, and by some coincidence the Ogg guys figure it out a year later, said Ogg guys are pretty much fucked.
posted by Talez at 10:10 PM on October 30, 2013 [3 favorites]

And, once again, this is why we need patent reform.
posted by Samizdata at 11:47 PM on October 30, 2013

You know what the sweet thing about published standards and patents? Unlike trademark and copyright patents have clear end dates...sort of.

I thought patents were done after 20 years, but it seems in some cases you can extend them? Ie there is an MP3 patent claimed to expire in 2017, when MP3 came out in 1993.

Anyway, assuming that according to Wikipedia the last h.264 patent runs out in 2027; however, I'm betting that is one of the newer bits, as the initial spec was written in 1999. So, why don't we just focus on the bits of the spec over 20 years old and make a free.264 version that implements only the parts old enough to be (probably) patent free?

Or an idea for the future; Churn out a bunch of formats, implementing every idea we can think of, then leave them for 20 years. Then in 2033 we've got a bunch of file formats ready to go.

Speaking of which; shouldn't the patents on JPEG 2000 be running out soon, since that was all done before 2000?
posted by Canageek at 11:55 PM on October 30, 2013

Man remember when the patents on mp3 totally held back digital culture and smothered music? Wasn't that terrible? We can't let that happen again, because the tyranny of MP3 destroyed music due to its patent encumbered nature. It would be terrible if digital video was as held back by codecs as music. I mean, all this time that Firefox and other open source software couldn't play music? Ridiculous! I'm glad that Firefix was willing to die on the battlefield over h.264 in the past, but shame on them for givng in to the destruction of culture by letting these patents anywhere near Firefox.

Seriously, the rhetoric around h.264 patents isn't in any way commensurate with the observed effects. There are terrible software patents, but these are not terrible software patents. And pretending that they are onerous just poisons the well for arguing against the truly terrible software patents.

Good on Cisco for doing this. If you want source, go compile the GPL x264 project or other implementations, which will hopefully include Cisco's soon.
posted by Llama-Lime at 12:33 AM on October 31, 2013 [1 favorite]

Llama-Lime: in case you weren't aware, Firefox still can't play MP3 audio unless the underlying platform includes a codec, precisely because of software patents. Despite being patent-encumbered, MP3 became popular because Fraunhofer waited until it was already entrenched before deciding to charge license fees for the decoder.

Open-source projects like VLC and ffmpeg can only use patent-encumbered codecs by willfully violating the applicable patents and hoping they won't be sued into oblivion. Just like Microsoft has little incentive to shut down piracy of Windows, groups like MPEG-LA don't really care if you violate their patents... until you start doing something that makes money, at which point they can bring the hammer down. You may think that's a perfectly reasonable situation, but it's completely antithetical to the free software ethos.

And as far as I can tell, compiling x264 and using it in a product is still just as illegal as ever, at least in the US.
posted by teraflop at 1:27 AM on October 31, 2013 [17 favorites]

So, does this mean I can encode video in one format/codec, stick it in an HTML <video> element, and it will play okay on Firefox, Internet Explorer 10+, Chrome, and Safari? Desktop and mobile versions?

What's the format, MP4 (example.mp4) containing video using the H.264 codec and audio using the AAC or MP3 codecs? (from Media formats supported by the HTML audio and video elements)? Or something else?

This doesn't help with <audio>, does it? Except I guess we'll all be making 1px videos and using the <video> element instead to provide audio?
posted by alasdair at 1:57 AM on October 31, 2013

rms does not approve
posted by mikelieman at 3:00 AM on October 31, 2013 [2 favorites]

You know what the sweet thing about published standards and patents? Unlike trademark and copyright patents have clear end dates...sort of.

Though, like all things, this can be hacked. One practice is to patent other things the original patent depends on at long intervals, forming a “patent hedge”
posted by acb at 3:20 AM on October 31, 2013

If browsers would hurry up and get faster we could all just run the JavaScript H.264 decoder. :)
posted by deathpanels at 3:35 AM on October 31, 2013 [1 favorite]

Comments from Monty Montgomery from Xiph. Says this doesn't cover the audio codec in a video, so I'm not sure what that means - you can do video with H.264 video but not AAC/MP3 audio? And it only applies to WebRTC, not more generally (as the post says) so the answer to my questions above is "no, this doesn't help with the HTML5 <video> element"?
posted by alasdair at 3:42 AM on October 31, 2013

Yes, AFAIK there are still no audio codecs that are supported by all major browsers on all platforms.
posted by teraflop at 4:00 AM on October 31, 2013

Isn't HEVC (H.265) where things are going anyway?
posted by Bathtub Bobsled at 5:34 AM on October 31, 2013

I'm pulling for HVAC. It's cooler technology, and the hot thing.
posted by snuffleupagus at 5:51 AM on October 31, 2013 [8 favorites]

Well, Xiph may have retreated, but I have a sneaky hunch that good ol' rms will still be cantankerous as ever.

So - the binary only - that I *am* worried about. I mean. Now that the whole NSA thing happened how the everloving hell do I have any guarantee that there isn't a backdoor in this binary module? I suppose people will use tools to snoop packets and see if there's anything fishy, I dunno. I just don't trust modern American tech companies who give a "free" lunch, anymore.
posted by symbioid at 6:48 AM on October 31, 2013 [1 favorite]

symbioid, AIUI there is also "open" source; while the patent restrictions mean you may not be permitted to modify or redistribute, it might eventually be possible to do your own build and confirm that the codec binary shipped by cisco matches the sources. (I say eventually because repeatable builds are a real bitch and I don't know if cisco is motivated to help make this happen by e.g., fully build-server configurations)
posted by jepler at 6:58 AM on October 31, 2013

And of course as an individual you'd be very likely to "get away with" building your own from source.
posted by jepler at 6:59 AM on October 31, 2013

I thought patents were done after 20 years, but it seems in some cases you can extend them?

In the US the rule is 20 years after filing, with the possibility of some term extension (computed at the time of issue) for delays caused by the Patent Office. This is rarely more than 3 years. However, before we switched to that rule in 1995 the old term was 17 years after the date of issue. Some applications took many, many years to be processed, and some others were filed after 1995 but can claim priority to an original application filed before 1995. So there are still a number of patents in force that use the old term.

But once a patent has been issued, there is essentially no way to extend its term.
posted by jedicus at 7:41 AM on October 31, 2013

Isn't HEVC (H.265) where things are going anyway?

In a few years... My understanding is that there's no encoder that's really ready for prime time at the moment, so your nice licensed decoder would have nothing to play. H.264 is the now and the near future.
posted by monocyte at 7:55 AM on October 31, 2013

I wonder if they're thinking of how the medical world puts forth a new use on a pharma patent for a drug, so it's "extended" (i.e. antidepressant goes to relabel/extend for PTSD, etc... keeps it off the market for generic company's production while they retain the exclusive right)
posted by symbioid at 7:57 AM on October 31, 2013

how the medical world puts forth a new use on a pharma patent for a drug

A new use for an old compound is not patentable. What can be patented is a new method of treatment, but those are hard to enforce. A generic drug company can still make the old drug for the original condition, which doctors can then prescribe off-label. The patent owners only option would be to sue the doctors themselves for patent infringement, which never goes over well.

What has been done is to figure out which enantiomer of a drug is (more) effective or safer and then get a patent on just that compound rather than the racemic mixture. An example of that is Nexium (esomeprazole) being patented over Prilosec (omeprazole).
posted by jedicus at 10:54 AM on October 31, 2013

MPEG LA and other technology groups have patent portfolios that are wide-ranging and obfuscated.

What do you mean by obfuscated? Lists of all the patents contained in the MPEG LA pools are publicly available.
posted by jedicus at 10:56 AM on October 31, 2013

Here's the one for h.264, for example [pdf].
posted by jedicus at 10:56 AM on October 31, 2013

See Forbes (no less):

That's just a blog hosted by Forbes. It doesn't represent the position of Forbes itself, which these days is basically a blog hosting service à la the Huffington Post more than it is a self-sufficient publication.
posted by jedicus at 10:58 AM on October 31, 2013

shouldn't the patents on JPEG 2000 be running out soon, since that was all done before 2000?

The patents covering JPEG 2000 held by the members of JPEG were already freely licensed for "implementable in their baseline form without payment of royalty and license fees" (source). Some companies (Forgent and Lizardtech) made some noise about patents they owned covering JPEG 2000 but in Forgent's case I think nothing came of it (though it was successful in suing over regular JPEG) and in Lizardtech's case it sued but ultimately lost.
posted by jedicus at 11:10 AM on October 31, 2013

acb: Though, like all things, this can be hacked. One practice is to patent other things the original patent depends on at long intervals, forming a “patent hedge” Wouldn't the original standard count as prior art then (Since anything it uses is obviously known....)

jedicus: So if we looked up the maximum delay at the patent office that year, we could tell when the last patent that could cover the format would be, and thus find when the last possible torpedo patent runs out?

Also; A lot of the patents are publicly known. However, if it is like drug companies, they will have additional, hard to understand patents that hide what they are actually about. So they will patent huge swaths around the actual drug, so their competitors don't know what they are working on (and to prevent someone from making a molecule that is a tiny bit different but works the same, as happened with viagra/cialis)
posted by Canageek at 2:15 PM on October 31, 2013

jedicus, I mean that the language in the patent is written specifically to confuse patent examiners and allow for unknowing infringement. See this wonderful link from Joel Spolsky.
posted by infinitewindow at 2:57 PM on October 31, 2013

So if we looked up the maximum delay at the patent office that year, we could tell when the last patent that could cover the format would be, and thus find when the last possible torpedo patent runs out?

Also I'm not sure what you mean by torpedo patent. Did you mean submarine patent? Those are basically a non-issue today. There are fewer than 600 pre-1995 patent applications that are still pending, most of which were subject to secrecy orders because they cover a classified technology.

So they will patent huge swaths around the actual drug, so their competitors don't know what they are working on

In my experience that has less to do with obfuscation and more to do with trying to patent as broad a class of drug as possible. But those patents and almost always have a series of narrower and narrower claims, typically ending in one or more claims covering only a single substance (typically the actual drug that was tested). For example, here's a patent covering Viagra (sildenafil). Claims 1, 7, and 8 are extremely broad, covering large classes of compounds. But claims 3, 4, and 11-19 are to specific compounds.
posted by jedicus at 6:14 PM on October 31, 2013