Web Video
Summary:
Issues:
- RF video codec adoption by W3C Goal: encourage W3C do a patent search & have a talk now or shut-up call for patents
- API inconsistency of Web Theora usage
Goals:
- encourage mv_embed adoption in CMSs
- research javascript API inconsistencies and provide feedback to WHAT WG as input to standards
Detailed notes:
CMS Systems
mdale: gonna try and get funding from Mozilla to put mv_embed into CMSs - plony, wordpress, drupal, et al
tim: but have you asked yet?
mdale: No :) Looking around for funding in general. Transmission network interested in committing to it. Later versions of mv_embed allow you to select which plugin which they like.
silvia: there's already some ogg embeddings for e.g. plone
shane: but mv_embed has more features
mdale: *gives demonstration*
- choice of plugins - extraction of embed links for sharing - etc.
silvia: had a discussion at the W3C video workshop dec last year. Heaps of people there concerning many different aspects. Started with groups talking about requirements, then technical talks about how to get hyperlinks into video, SMIL presentation, MPEG4 presentation - all criticized for only addressing into XML documents rather than video itself. I explained that you really need a server component in order to do this and that opened a few eyes. Key: can't just look at one aspect. Hyperlinks -> server compenents -> video segmentation -> etc. etc. Need to look at the problem as a whole rather than addressing individual components. Then discussion about what video format to have.
Open source point of view: we want to know about any dangerous patents because then we at least have the chance of working around them. Stefan from Nokia thinks there may be less patent issues in the wavelet space so Dirac might be better choice.
Tim: I would have thought the opposite was true.
Shane: I agree - many of the predictive patents are for MPEG4 and AVC which came out after Theora, not before.
Tim: and a whole bunch more is expiring soon or has already expired. Wavelets it currently getting very heavily patented and these may have a bearing on Dirac.
Rob: Has always been a minority group in MPEG that wants to make MPEG royalty free. Sun, Microsoft, Telcos, Chinese companies, some others. Chinese companies left en masse and created their own codecs as a result, then got threatened by consumer electronics companies into not releasing them royalty free.
Preliminary versions of AVS H.264 were presented to be royalty free, but these got voted down by specification committees. MPEG4 pool was so expensive that it was imploding and they needed a new less expensive version - hence H.264.
Conclusion: very unlikely that MPEG committee or any other standards committee will release royalty free video codecs. Also very unlikely that participating companies don't know about patents of interest. Pretending there are unknown patents around helps create uncertainty which is in their interest as it reduces the risk of royalty free codecs arising elsewhere. VP-3 is probably based on H.261 and H.263 which don't have patent pools - there are plenty of these around (e.g. sorenson, real) and none have been knocked out of the market by litigation. Also some problems with holding back a patent instead of speaking up when you first notice infringement - submarining tends to lead to patents being revoked these days.
Andre: so is a patent search useful? and what else should be done
Silvia: Patent search and documentation can lead to agreement that all believe there's no patents around that theora infringes on. Then W3C could get up and start requesting patent holders to speak up.
Rob: W3C has a tradition of doing this.
Rob: There's patents around everything, there's patents around the DCT, around wavelets. But Sun has said internally, if anyone ever wants to sue us over the DCT then bring it on. Each patent covers a very, very narrow aspect of the DCT. And one of the advantages of using mature technology is that there are generally very few large blockers.
Rob: Is the W3C going to decide something on the video image?
Silvia: They want to, if they have to go back to use a 20 year-old format, they will, though they don't want to do that.
MikeS: They could just not specify anything.
Silvia: They've learned from other formats that that's not a good solution to achieve interoperability.
MikeS: The real danger of specifying something like H.261 is that it is not at all competitive, so everyone will implement something else as well, fragmenting the market.
Rob: Seems to be all these shadowy claims that are rubbish on close inspection (can't improve codecs, no-one knows what's in VP-6). So of course it's possible to build new royalty-free codecs.
API discussion:
mdale: there's stuff missing from WHAT-WG's video element but a lot of things can and probably should be done in javascript (e.g. playlists)
Summary:
Issues:
- claims that Theora has patent risks
- API inconsistencies & problems. Even our API is incomplete but the HTML5 / Cortado ones are missing a lot of features.
Goals:
- get W3C to do patent investigation on Theora then make announcement that any who believe there's patent risks should stand up.
- provide input to WHAT-WG about HTML5 video elements Javascript API by analysing existing APIs (Annodex, VLC, Cortado, etc.).
Melbourne at night - Photo by mugley