Dan Dennedy: Kino and MLT Developer
Saturday, 06 March 2021
Main Menu

cross-platform video editor
Media processing framework
Kino and dvgrab
Non-linear DV video editor for Linux
Linux 1394
Firewire drivers for Linux

Syndication Feeds
RSS 2.0 Feed

Dan Dennedy's Facebook profile

Video Codec Patents and Ogg Theora  
Friday, 07 March 2008

There has been a lot of discussions in the past few months about patents on video codecs and how Ogg Theora fits into the picture. Most recently I read Ralph Giles' blog post, and I feel the need to help clarify the concerns about Ogg Theora because I do not think he quite grasps it yet. Or, maybe he does, but he just does not accept it, placing higher value on different points.

Modern video codecs are complex and implement quite a few concepts. There are patents on most of the algorithms implementing these concepts. During the standardization process of MPEG-2 and MPEG-4, the majority of the patents are revealed, and then the patent holders establish a means for paying a royalty/fee for a right grant on the whole pool of patents. On the other hand, On2 only made VP3 available royalty free for their own patents. Considering how Theora is not revolutionary and implements a lot of the same concepts used in ITU and MPEG codecs, there is a good chance some other parties hold patents on some of these algorithms.

I do not describe these as "submarine "patents. This is not about submarines, and that term is being used in a loaded manner of tone. It is a simple matter of facts, probability, and risk. Considering the nature of the technology, there are many things that can be covered by a patent. While there is still some risk with the MPEG codecs (there is always some risk), there is a low probability that something is not covered by the patent license pool. With VP3 and Theora, only some smaller subset is covered by On2 patents.

Yes, the software patent situation does suck. What sucks most is perhaps the uncertainty and fear of litigation. So, yes, this is a matter of FUD and for good reason. To me, it makes more sense to choose something with a known risk that is easily mitigated. I do not think we can simply feel safe just because On2 says we are safe from them. So, what Theora patents not claimed by On2 am I talking about? I do not know, but considering it is a rather traditional video codec, I'd hazard a guess that it's a decent amount, and they may not surface until they near their expiration.

When you combine this risk with Theora's lackluster quality, I just do not see a compelling reason to advocate or use it. On the other hand, Dirac and Schrodinger seems like a great way forward! This is a true win-win situation because, for one, BBC forgoes its patents and with it being a less traditional codec, it has a smaller area of exposure thereby reducing its risk. Additionally, you do not have to financially reward the crummy software patent system.