Yesterday I had a random technology developer email me with the question why he should use Ogg over other codecs that have a much more widespread uptake. Of course with “Ogg” he meant “Xiph codecs”, since a comparison of container formats isn’t really what people are asking for. He felt positive towards open codecs, but didn’t really know how to express this with reason. So I wrote up some arguments that can be made for open codecs.
First of all the royalty-free character of Xiph technology makes it possible for them to be used for any application without having to consider what impact the use of the codec has on ROI and scalability of business models. It is important to have a video and audio codec available that you can just use for exchanging audio and video data
just like you can exchange text – nobody would consider paying a license for ASCII either.
Second the flexibility that you get with Xiph is important for developing new applications. One example is the development of a scheme for encrypting audio or video for DRM and then transport it inside Ogg. Since everything around Ogg is open, one can just go ahead and implement this, even if the Xiph community is not interested in such technology.
Third let’s talk about quality. Ogg Vorbis is an audio codec that is of higher quality at comparable compression rates than MP3. Ogg Theora compares well to H.261 and also to MPEG-2 video. To achieve high-quality video such as in H.264, you will need to move into Dirac territory. And yet, Xiph has more to offer: for VoIP you can currently use the highly competitive Speex codec. And a new, hybrid speech/audio codec of low delay and high compression rate with very low quality loss is CELT, the new codec developed by the author of Speex. CELT has no comparison in proprietary codecs. All of the software is available for free and in source code from svn.xiph.org and the authors are easily reachable for discussions. Should there be need for improvement, everyone has the opportunity to develop such.
Lastly, I’d like to look at the capabilities of Ogg based technology in a Web environment. Over the last years we have developed technology that is now being included by the W3C into future Web standards. This includes URL addressing to time offsets into videos, which can ultimately help develop e.g. a Web-based video editor. This includes the implementation of Ogg Theora/Vorbis as baseline video codec in Firefox for the new HTML5 video element. This includes technology for making audio and video accessible, in particular to Web search engines through deeply searchable content, but also to hearing- and visually impaired. All the base techology and specifications are available.
“Ogg” is totally the future of media technology, because the future has to be open and royalty-free to allow everybody on this planet equal rights and possibilities to participate in a media-centric Internet, and because anything else will continue to be a burden on innovation.
Great post.
Are there much examples of commercial usage of Dirac?
It seems to be very little development/discussion about Dirac these days?
Rex.
Dirac was only just released as version 1.0 in September 2.0. It is now stable and usable and has two independent but interoperable implementations. It is also standardised through the SMPTE as VC-2 and starts being built into hardware, e.g. http://www.broadcastwarehouse.com/avt/titan-vc2-video-codec/1157/product . While I haven’t seen large collections of Dirac files yet (anybody?), there is certainly lots happening around it. You can get the latest at http://sourceforge.net/projects/dirac/.
http://diracvideo.org/ is a better link for dirac-related resources.
Hi, thanks for writing… I’ve been trying to find a rationale for using those old codecs.
The basis of the argument seems to be “so that people can make new tools without paying for the use of modern codecs”, is that correct?
If so, then do actual content producers or viewing audiences come into this equation, or is it just directed at tool producers?
(btw, I agree on the general desirability of widespread support for patent-unencumber codecs. I object to the “proprietary” bashing that often goes along with it, but suspect we’re more in alignment on general goals than in actual opposition.)
tx, jd/adobe
John, thanks for posting – I’m sure there are more people out there who have similar impressions to yours.
So let me start by saying that while the un-reflected bashing of “proprietary” is indeed an issue with “open” enthusiasts, the bashing of “old codecs” is an issue with “proprietary” enthusiasts. Vorbis is a newer codec than MP3. Dirac is a newer codec than H.264. CELT is a newer codec than anything else. The only open codec that can really bear the sign of “old” is VP3, which is the basis of Theora – but even Theora has developed heaps since VP3. So, maybe we can do away with the “old” stigma on open codecs.
Now let’s look at the rational for patent unencumbered and royalty-free codecs.
As I mentioned, I see the use of proprietary and patent encumbered codecs as an inhibitor towards innovation. This is indeed a problem for tool developers, who are restricted in their abilities to develop new functionality around codecs because the codecs are not accessible. Yet, restricted tools reflect on restricted functionality of applications, and thus restricted possibilities for content producers and audiences. So, this reflects on the complete value chain of audio and video.
Further, I see a need for royalty-free codecs in certain application areas. Don’t get me wrong: in certain business areas I don’t see a problem with licensed codecs – there is enough money made that the royalties for the codecs don’t have a negative impact on the businesses that use them. But when it comes to the Web in particular, the basic data types for publishing need to be available freely, such that the majority of Web publishers – in particular private people and non-profits – are able to publish just like everybody else. Here, “free speech” actually depends on freely available tools.
Since for typical licensable codecs, there are royalties payable on encoders, decoders, and content distribution, basically everybody in the value chain of audio or video will have to pay eventually, not just the tool developers.
So, the answer to your question about whether content producers and viewing audiences come into this equation is an absolute yes – both from a licensing point of view, as well as from a restricted functionality point of view.
Hope this helps explain some more of the motivation behind Xiph’s codecs.
Thanks. So, I’m understanding the argument as “Use of older patent-unencumbered codecs will help remove one of the barriers for toolmakers, and this will then have benefits for publishers and audiences”… am I getting that correct?
(Two of the big ignored questions since the start of VIDEO tag discussions has been how to move publisher workflows into a different codec, and then how to earn enough consumer adoption to make it worth publishers’ effort.)
I agree with you that “everyone pays for codec innovation”, because when companies like Adobe foot the bill for updating the world’s desktops and devices, that’s an opportunity cost for other things Adobe could have done with the money. On the other hand, the techincal improvements offered by modern codecs offer benefits of their own. Fortunately, the cost of modern codecs is not paid directly by either content creators or audiences.
tx, jd/adobe
Sorry for the late reply.
John, you asked how to move publisher workflows into a different codec in view of the possibility of HTML5 choosing Theora as the baseline codec. I agree that this is a hard problem.
A person that has an existing workflow of video production and publication will not want to change that. And indeed, up until the last step where it goes to publishing, they probably won’t even if a new codec is in town. Even if I publish my videos in Ogg Theora format, I will continue using iMovie or Windows Movie Maker and probably the DV format for editing.
An organisation that has an existing transcoding workflow will actually not have much of a problem. They will simply replace or add to their existing workflow a transcoding to Ogg Theora, which can be easily done using ffmpeg2theora.
People that are new to the HTML5 Web and would like to author videos may indeed be the most interesting of all. Not only will they be keen to use online tools to edit their video. They will also set up new workflows that will include transcoding and handling of Ogg Theora video. It’s this group of people that tools other than transcoding tools are directed towards. It’s these people that will appreciate the new tools online around Ogg Theora, e.g. Metavid, or Firefogg. More tools will certainly be necessary to address the full set of issues.