[whatwg] on codecs in a 'video' tag.

[Maciej Stachowiak]

On Apr 2, 2007, at 5:03 AM, Gervase Markham wrote:

> Maciej Stachowiak wrote:
>> Reasons Apple would like MPEG4 + H.264 + AAC to be the preferred  
>> codec stack
>> ----------
>> - We already need to support these for video production and  
>> consumer electronics (so no extra patent cost to us)
>
> I don't understand this point. There's no extra patent cost in  
> supporting Theora. (See below for submarine patents.)

What I mean is that unlike the case for other browser vendors, it  
won't cost us anything in patent license fees.

>
>> - Every extra codec we ship is incrementally more submarine patent  
>> risk (which could cost us hundreds of millions or billions of  
>> dollars)
>
> But this is not just true of video codecs. Is Apple planning to  
> stop shipping new software and improvements in Mac OS X because  
> some of it may be patented?
>
> If you are concerned about submarine patents, I suggest that "not  
> shipping stuff" is not a sustainable strategy to counter them.

Obviously there is a tradeoff. Video codecs are a much more patent- 
prone area than many other areas of software.

>> - They are technically superior to Ogg (seekable container format,  
>> significantly better bitrate for video)
> > - They are competitive with likely next-generation proprietary video
> > formats
>
> I'll let others comment on this. But I would note that JPEG2000 is  
> technically superior to JPEG, but hasn't been widely implemented  
> due to patent issues.

And due to the fact that JPEG is already universally deployed and  
good enough for most uses.

>
>> - They are an open ISO standard (patents notwithstanding)
>> - They are widely available in hardware implementations which we  
>> can use in our Consumer Electronics devices
>> - They have been chosen as a standard for 3G mobile devices, HD- 
>> DVD, Blu-Ray, HDTV broadcast, etc
>
> All of which ship in countable units, and (where applicable) don't  
> run free software.
>
>> Reasons Mozilla would like Ogg + Theora + Vorbis to be the  
>> preferred codec stack
>> ----------
>> - All known patents are royalty-free, so no need to pay $5 million  
>> to MPEG-LA
>
> The problem is not that it's $5 million, it's that the amount is  
> unknown and unmeasurable. They have no "fixed fee above a certain  
> number of units" licensing policy. And even if they did, a Mozilla  
> license wouldn't cover other members of that community.

Actually, they do have a license cap, and I overestated it. See  
<http://www.mpegla.com/m4v/m4v-agreement.cfm>. It's only $1 million  
for "Decoders sold to end-users and/or as fully functioning for  
PCs." (This document could be out of date.)

It's not immediately clear to me that a Mozilla license would not  
cover redistribution, for instance the license fees paid by OS  
vendors generally cover redistribution when the OS is bundled with a  
PC. I think someone would have to look at the legal language of the  
agreement to see if it covers redistribution.

>> - Implementation would clearly be freely redistributable by third  
>> parties (the situation might be unclear if only Mozilla paid for a  
>> patent license)
>> - No demand for use fees for commercial distribution in this format.
>
> Let me add other reasons why Mozilla (for whom, again, I am not  
> speaking) might want to specify Theora/Dirac:
>
> - They have a strong commitment to interoperability

I don't think Theora (or Dirac) are inherently more interoperable  
than other codecs. There's only one implementation of each so far, so  
there's actually less proof of this than for other codecs.

> - They appreciate that there are a wide variety of distribution  
> models;
>   for browsers, and do not want to choose technologies which work only
>   for some of those;

Unfortunately, Ogg does not work for some browsers either.

> - If they think a royalty-free patent policy for standards is a good
>   idea in one place (the W3C) then they think it's a good idea
>   everywhere.

The problem is that the main standards bodies for video (such as the  
ISO) do not have the same norms about RF vs. RAND patent licensing as  
the W3C.


>
>> We think your reasons are strong and worthy of respect. That is  
>> why we are not trying to force our codec preference on you, but  
>> rather propose to leave this issue open. We ask you to respect our  
>> reasons as well, rather than trying to force us to go along with  
>> your codec preference.
> >
>> I think achieving broader interoperability will require us to find  
>> ways around this impasse, rather than bludgeoning each other until  
>> one side caves.
>
> So, just to be clear: you believe interoperability is best promoted  
> by having no codec specified in the spec?

I think if the spec mandates a single codec, that part of the spec  
will be ignored by at least some parties.

>
>> One possibility would be an open API for codec plugins that will  
>> work in <video>/<audio>, then user availability of codecs is not  
>> directly tied to browser choice and codecs can compete in the  
>> marketplace more freely.
>
> You and I both know that this would result in dominance for  
> whatever codecs got shipped by default on major operating systems.  
> Content producers will not choose codecs for 5 or 10% better  
> quality or bitrate, they will choose them for user convenience -  
> because if their site is harder to use than their competitors,  
> they'll fail.

Isn't this basically admitting that Ogg Theora would fail in the  
market if not legislated in the spec? Still, I would not be so sure  
of your conclusion. The bitrate differences among current codecs have  
a range of 2x or 4x, not just 10%. Also, Mozilla supporting Ogg would  
probably carry more weight than Safari supporting Mac OS X, since  
default or not, it has greater overall use share.

> As codecs are binary components, the site wanting to use foo-codec  
> would need to provide versions of it for every operating system  
> they planned to support. For Linux, that would be rather  
> complicated, to say the least. They might not even bother offering  
> it for Mac, or Mac PPC. This would discriminate against operating  
> systems with smaller market shares.
>
>> Another possibility would be to get MPEG-LA to change licensing  
>> terms somehow.
>
> I'm sure that any help Apple would be able to give in this area  
> would be much appreciated. How do you suggest we begin?

One good first step might be for someone to obtain a copy of the  
existing license terms and determine how they would apply to a freely  
redistributable product.

>
>> Yet another possibility is that one codec stack will become so  
>> popular that all parties will feel compelled to implement it  
>> despite their reasons against.
>
> You again assume that only recalcitrance prevents some parties  
> implementing any particular codec stack. As I understand the  
> situation, Firefox would have to stop being free software in order  
> to ship an MPEG4 implementation.

I don't think that is true, but it would depend on the details of the  
MPEG-LA license agreement. Also, at most the MPEG4 implementation  
would not be free software, this would not have to affect the rest of  
Firefox.

Regards,
Maciej