- From: Håkon Wium Lie <howcome@opera.com>
- Date: Fri, 6 Apr 2007 01:29:34 +0200
- To: Chris Wilson <Chris.Wilson@microsoft.com>
- Cc: public-html@w3.org
Chris Wilson wrote: > > We would personally prefer the MPEG-4 family of codecs (AAC, MPEG-4 > > Part II, H.264) to be a common baseline. > > We don't think it's appropriate to mandate it in the spec, though > Yeah, what he said. :) Let me take a different view. I will argue that (a) having a universally understood baseline video format is a good thing for the web; (b) that in order to achieve this, the specifications must mandate it; and (c) that MPEG-4/H.264 isn't it. (a) is good for the sake of interoperability, and few will argue against having a baseline format. Some will say, however, argue against (b). It has been pointed out that the HTML specification didn't mandate specific image formats, but that we still ended up with GIF and JPEG being universally understood. The "market" sorted it out, it is said. I disagree. Marc Andreessen sorted it out when he added support for <IMG> in 1993. He added support for GIF an JPEG and all browsers that came after him had to support these formats as well. So, for image formats, it wasn't necessary to mandate support in HTML, the choice of formats had already been taken by the time specifications were written. This is not the case for video. Now to (c). MPEG-4/H.264 has been suggested as being the universally understood format we are looking for. It's not a good choice, either. First and foremost, it's not royalty-free. MPEG LA and others (Qualcomm, it seems) are demanding royalties from those who implement it or make content available in it. Therefore, H.264 is not royalty-free (RF). It's a pronounced goal for W3C to ensure that the standards of the web are RF. We want everyone to produce and use video on the web, not just those who can pay the dues. For example, OLPC's $100 laptop has a video camera but it would be immoral to raise the price to pay MPEG LA royalties. So, what can we use instead? Personally, I think Ogg Theora is a good choice. Wikipedia has chosen it as their video format, and hundreds of snippets are available: http://commons.wikimedia.org/wiki/Category:Video Dirac has also been mentioned as a promising RF video format. If any of you have other suitable candidates, it's a good time to step forward. Perhaps one of the levels of H.264 isn't royalty-encumbered after all? You can find an experimental version of Opera that has native support for Ogg Theora and (an early version of) the proposed <video> element. We invite you to play with it: http://people.opera.com/howcome/2007/video/ The <video> element is described here: http://www.whatwg.org/specs/web-apps/current-work/#video Cheers, -h&kon Håkon Wium Lie CTO °þe®ª howcome@opera.com http://people.opera.com/howcome
Received on Thursday, 5 April 2007 23:29:55 UTC