As announced on wikimedia late yesterday, Mozilla will be investing 100k into Theora development over the next 6 months. This development effort includes 3 projects aiming to rapidly mature core open video technologies for wider use.
The three project include the following developers:
Viktor Gal - the maintainer of liboggplay
Conrad Parker - the key developer of multiple Ogg support libraries, in particular liboggz
Tim Terriberry - the key developer of Ogg Theora
Viktor will work towards stabilizing the current Ogg Theora support via the liboggplay library in Firefox and (time willing) add Schrödinger dirac support. Conrad will work on improving Ogg network seeking, language selection and improved core library support. Tim will work on encoder enhancements to enable the new Thusnelda Theora encoder. These projects combined will make free license video very competitive with contemporary codecs & containers and will enable free video to greatly surpass contemporary solutions in terms of scalable web delivery.
In my LCA talk I outlined how open media is becoming very competitive as well as the latest development updates for collaborative video on wikipedia in partnership with kaltura. I will try and post a summary of that talk shortly.
Here is a summary of FOMS 2009 discussions and developments that I see are applicable to collaborative media on Wikimedia’s sites.
FOMS 2009 included a lot good discussion and hacking. As with previous years the meeting included the establishment of Community Goals
Several Projects are of interest to the future of collaborative media on Wikimeida.
Software Patents
As you may remember last December Nokia raised some “patten issues” to get the free format ogg theora tossed out of the html5 spec. While this failed to stop firefox from shipping ogg it has slowed down Opera and given Apple a reason not to ship Theora support. OpenMediaNow.org is project attempting to take on patent issues around enabling media in free software in a legally conforming way. They aim to encourage developers and companies to publish research around open codecs and group together legal resources to confront this issue.
Client Encoding
As mentioned on wikitech-l, firefogg is a really great solution for uploading theora media to websites from within the web browser. You simply point it at your high quality HD or DV footage and it transcodes to theora from settings supplied by the web service and uploads (in this case commons). Some new features for firefogg where discussed including resumable uploads.
Server Side Flattening of Sequences & Video editor
The current state of PiTiVi the open source gstreamer based video editor was presented. PiTiVi has been a 5 year effort Edward Hervey that has some very nice distinguishing qualities over the other open source video editor efforts. Most importantly its a extremely modular gNonLib / gstreamer core. (allowing it to easily tie into the massive collection of gstreamer plugins). PiTiVi’s is written in python and its interface is optional. We discussed how it could serve as a server side flattener for video collaboratively created on Wikipedia. This way you could save a sequence to a DVD or play back the flattened sequence with the java applet or a video plugin (just like video files are currently supported on Wikipedia).
Oggz_Chop
The current state of oggz_chop was reviewed and a model for highly scalable, accurate media seeking and playing via integrated open protocol for client side (liboggplay, firefox) and server side ( oggz_chop) ogg serving. Essentially oggz_chop provides headers for Firefox so it can request two resources one small oggz_chop generated resource ogg header and other being a normal http byte range request for the video payload. This will work well with proxies that already support http ranged requests. If your client does not send headers saying it support this model you will continue to get a single resource so things like wget myvideo.ogg?t=start&end will still work
Bright Future for Open Media Scaling Up & Support for Dirac
Diracsupport was also discussed. Unlike the many years of efforts to get theora out into the media ecosystem we will likely be able to push Dirac out a _lot_ faster. This is because free code libraries are now powering media for popular web browser. We discussed adding Dirac support to both Firefogg (the in browser transcoder) and liboggplay (in library for supporting media playback in Firefox). Both efforts are relatively far along as ffmpeg2dirac already exists and some patches for liboggplay are on the way
Archive.org has undergone a massive transcode effort to make all their media available in the free & open format ogg theora.
This massive transcode effort by Archive.org will greatly expand interoperability with Wikimedia hosted projects and the metavid’s archive. Metavid can now use Archive.org as a transparent hosting provider for Metavid videos ensuring we can scale up with increased usage. For Wikimedia, this will make it easier for Wikipedian’s to work with Archive.orgs huge collection of public domain material. Archive.org moving images collections contains hundreds of thousands of digital movies which range from classic full-length films, to daily alternative news broadcasts, legislative proceedings, and a great deal of public domain & free content licensed video. This Interoperability with Wikimedia commons makes it easier to use archive.org content to enhance Wikipedia articles.
Server side segmentation technologies are also being deployed to improve inter-archive interoperability with open temporal media formats. The Annodex foundation a sister organization to Xiph.org, focuses on timed text, segment video reference, and core software libraries to support video playback in the Firefox browser. Annodex server side tool oggz_chop which has been deployed on archive.org allows for timed video reference. For example requesting this clip:
gives you a 10 seconds from 20 minutes into the film of the Night of the Living Dead. You can download that piece and upload it to Wikimedia. (or an interface could let you visually set in and out points and seamlessly reference that segment) This is essentially the same annodex server side segmentation technology we have run here on metavid for years. We are working on integrating these open tools for inter-archive interpoerability tools on Wikimedia as well using these technologies with web sequencer efforts in collaboration with kaltura.
The Metavid Archive has captured video and text captions from cable broadcasts of public domain house and senate footage since 2006. We have made all this content available for search and reuse in an entirely open source video platform. But the archive is not perfect, over-the-air cable broadcasts do not provide perfect sync for close captions, and live transcription is not 100% accurate.
We are now calling on visitors to share and promote the Metavid site in order to build a more perfect archive. Specifically, we are asking people to try and spend a few minutes of their time to try out the transcript improving tools. For more help on how the transcript improving process works, check out the improving transcript accuracy help section here on the wiki.
Improving transcript segments goes a long way here on metavid because the metavid platform builds on the most powerful collaborative knowledge production platforms in existence: mediaWiki (the software that powers wikipedia), and semantic mediaWiki a powerful structured data extension to mediaWiki. Your freedom to collaborative is protected by creative commons by-sa license ensuring your freedom to reuse the archive in its entirety for any purpose as long as you don’t prevent others from doing the same.
The More Perfect Archive We are Building
These same improved transcripts are carried over when people embed posts in blogs, enhancing the content accessibility. The transcript is exportable in the open cmml timed text format it can be muxed with the ogg stream for archival distribution and is easily searched as the text is directly in the page or accessible in machine readable CMML. (not hidden or encapsulated in a proprietary player like the approach of some flash subtitle sites) Annotative layers can categorize larger stream segments of video enhancing searchability and contextualization of media segments.
Since your participation in the metavid archive semantically tags time segments and we scrape information from a half dozen open congress sites; with your participation powerful semantic queries become possible.
The site give users powerful tools to create pages that highlight particular issues for example see the bailout coverage page, and provides endless mashup opportunities. We will continue to improve the archive as we make edits to transcripts. We will continue improve the underling open source software and hopefully lay the groundwork for future collaborative video archive projects. How perfect an archive metavid becomes is only dependent on our imagination and collective participation.
We have rolled out massive feature updates we have developed, the new skin and interface features developed in collaboration with the participatory culture foundation and Summer of code contributions by Stjepan Rajko. Simultaneously we are doing 1.0 release candidates for the metavid software packages: the complete metavidWiki mediaWiki extension and the stand alone html5 javascript embedding library mv_embed. This post summarizes the software updates, exciting development underway in collaboration with kaltura & wikimedia foundation and outlines the up-and-coming launch for metavid.
Existing developments with Kaltura and Wikimedia
Work is under way to write a ~simple reduced feature set~ smil like player & editor using firefox 3.1 html5 video support. Basic transitions using css animations and frame scrubbing/serving features have been completed… watch this space & kalturas blog for updates as development moves forward. As this development matures we should have an integrated sequencer on-par with kalturas flash based video editor but using ogg theora and accessible to Firefox users on free software platforms.
Metavid Launch Plan
We will be launching metavid over the next week or so. Watch this blog for updates As a community archive project users help in reporting any bugs they find with the site in #metavid on irc.freenode.net or emailing them in to the developers list
MediaWiki rc1 release notes:
Some minor known issues remain please send in bug reports as we move towards an official 1.0 release
Unified Search
new unified search model groups and aggregates relevant semantic metadata per search
advanced search improvements
improved autocomplete suggestions/display
Stream Interface Enhancements
new streamlined stream interface for easier clip editing and jumping around the stream
improved transcript integration with video playback (scrolls transcript window while seeking and playing back content)
Improved Skinning Support
improved skin infrastructure to support the skin created by the Participatory Culture Foundation
Mv_Embed (1.0.rc1)
improved html5 -> current browsers/plugins support
improved transcript scrolling, remote embedding without data proxy, metadata queries, and browser compatibility.
improved video control skinning (should soon be possible to support multiple css based skins)
developed php based flash media server added for serving segments of flvs to arbitrary clients.
(based on FLV4PHP this lets you do server side media seeking similar to other php based falsh-media-servers except mvflvServer its compatible other playback clients like vlc and mplayer, works with flv files over 4 hours 39 min 37 seconds long and avoids sending the client keyframe metdata for quick streaming response)
As you may have heard kaltura will be sponsoring the development of open source video integration features for wikimedia projects. The news is covered in wikimedias blog:
Michael will work on adding support for video editing operations and other video-related functionality to MediaWiki, with a rich user interface built entirely on open standards like Ogg Theora. Michael’s work priorities will be coordinated between Kaltura and WMF. I am hoping that we can make incremental improvements to Wikimedia’s video capabilities that will start to become visible to users soon.
Hopefully someday soon we will be able to insert remixes of congress videos (among others) into wikipedia articles
Firefox 3 download day was a huge success and it features many improvement over firefox2. But as they say you can’t please everybody, and download issues were not the only blip on this otherwise exciting launch. Perhaps lost in the hoopla over Fierfox 3 impressive new features set is the html5 video support which did not make it into this release. While Chris Double has done an excellent job in building cross platform ogg theora support into Firefox the new implementation strategy raises some questions about the future vitality of open media and open web standards.
Specifically Mozilla current implementation strategy proposes supporting video via hooks into the proprietary media platforms for windows and mac. i.e Firefox on mac will hook into quicktime, Firefox on windows will hook into direct show, while Firefox in Linux will hook into gstreamer… This approach risks abandoning support for a baseline free codec (ie ogg theora) for the video tag. We can only hope the base cross platform theora support code that is already written is not abandoned as they add in these hooks.
Its now clear that html5 is heading in a codec agnostic direction, thanks to “patent concerns” raised by Nokia and apple pushing forward with quicktime based video tag “support” in Safari 3.1. This of course makes it complicated for sites like wikipedia to count on the html5 video tag to support free media since apple is not going to make any attempts to support it on their own. This makes it slightly more complicated for us at metavid as well. (will have to do some more case detection with the mv_embed script).
What is not clear is mozilla’s long term strategy….
Proprietary media formats do best in proprietary media platforms. Free software does poorly when parts of the web are tied to proprietary platforms. Free software can flourish on an open web. From a purely business perspective anything Mozilla can do to promote a web that works well with free software will ultimately increase the vitality of free software platforms and ultimately increase firefox’s market share. Apple and Microsoft are certainly not going to pre-install mozilla as the default browser in their proprietary platforms.
But forces beyond mozilla’s control have made htm5 codec agnostic… What should mozilla do now?
They should take the sonbird approach and ship gstreamer across all platforms with native wrappers to direct show and quicktime to fall back on proprietary encoded content while ensuring baseline free codec and free media platform support. There are many reasons to go with an open community extensible free media platform, as outlined in the songbird blog…but let me mention a few more:
Going with a GPL based media platform ensures they will not be subordinate to the same proprietary platforms that they are competing with. Ie Mozilla will control the media platform user experience rather than the host operating system media platform which has its own browser to promote.
It ensures that the widely successful Firefox extension concept can be extended to the media platform in a cross platform friendly way. There is already a huge base of gstreamer plugins to build off of. Imagine a site that installs a simple client extension that enables transcoding from local high quality DV format directly into dirac or theora and uploading to the site… avoiding server side transcodes and associated quality loss. OR an extension that lets you do live brodcasts from your machine integrated with an associated web service or a full featured inbrowser video editor OR an extension that uses gstreamer to enable output to DVD iso from a collaboratively produced video sequence –these sort of killer apps can help propagate firefox.
Doing gstreamer cross platform will make it easy to support future free codecs such as dirac in one pass, instead writing and maintaining codec extensions for every codec for each propitiatory platform and trying to ensure the experience is identical across both proprietary and free platforms.
The code is already mostly written.
If Mozilla takes its own manifesto seriously hopefully they will be more forward thinking about the open media platform issue.
Engage Media has published a very detailed report on the state of Free and Open source software for online video. Titled: FOSS Codecs For Online Video: Usability Uptake and Development 1.2 the paper details many software application and tools for ~free~ online publishing of video. They even include a mention of mv_embed ;). Their recommendations for pushing for ogg theora in HTML5 and using FOSS whenever possible in Transmission members projects are positive directions for wider open source media wide adoption. check it out
Here is a short demo (ogg 5 meg, mp4 17megs) video of the metavid wiki integration that has been under development. This video shows the stream view with basic editing of text and person attribute meta data. This is more or less what I demoed at wikipedia a while back. The in-development version of metavid wiki has 2 additional interfaces. One for searching streams metadata and another for creating and editing sequences. As was mentioned last week, the sequence playback functionality will be handled by mv_embed.
We hope to have something online that people can play with soon and then a total site conversion not long after that. To get a feel for how sequencing and “dynamic” searching will look/feel you can check out the original version of metavid… and then imagine in the mediaWiki platform ;P
Al Gore was on the Daily Show last night and had a good interview with Stewart. Gore criticizes the news/tv medium for its efforts to relate Congregational conversations to citizens. Television is a broken system that produces a one way conversation. Here is a snippet from the show.
Gore like others sees potential for improvement of these system of communication and governance via “The Internet”. In this respect Gore highlights the problems that we hope metavid will help address. Mainly making Congressional video more accessible and re-usable. We can imply by Gore criticism of television as a one way conversation that when he says “the Internet” he means the kind of Internet Lessig describes, a participatory read-write web. In software metaphors the existing television system is read only, closed source and proprietary. So if we aim to address this problem our social software design decisions should be informed by the ideals of free software and participatory culture. see metavid thesis
Gore goes on to say this Internet potential was not actualized in 2003. While true the� invasion, occupation and total devastation of Iraq was not prevented :(… more than a million people in the US and around 10 million people worldwide did hit the streets to try and stop it. This un-paralleled mass action was enabled in no small part by the improved social communication infrastructure. If we are to hail improvements in communication infrastructure as the savior of our democracy it would be good to highlight not only what “the media, the President, Congress, and CIA”, did wrong, but also what the people did right.
Engage Media has published a very 