*

I agree with your analysis that Ogg hasn’t seen the industry uptake yet that H.264 has. I am expecting this to be a matter of time, since not many commercial services around Ogg have been developed yet.

I believe in future Ogg will serve the lower end market – anyone wanting to host videos cheaply – and H.264 serving the upper end of the market – anyone in need of DRM and other functionality.

I am expecting government ultimately to be using Ogg, but there will be a transition period, where the lack of commercial uptake of Ogg support will simply slow down uptake of Ogg.

http://www.parliament.act.gov.au/live/outside.htm

The media is available in DoD within an hour after broadcasting, three times a day. We wanted to avoid transcoding during the workflow.

In our tests with a few different temporal techniques, RTSP won out for me because it’s a documented standard.

*

Your point about the MPEG-4 H.264 license is interesting and troubling for many. On the positive side, it’s everything we’ve ever wanted:

* standards-based
* widely deployed
* scalable from phone to broadcast
* efficient

However with an unending license fee for use of the standard within open video archives (EDU, GOV) it requires the licensee to assess their future usage (a simply impossible task for a nationwide, long-term service).

It would be good if there was a government-wide license for use of the MPEG standards, meaning that individual agencies would not have to deal with the problem.

The OGG alternative is a great contender for desktop use. But I’d contend that it compares poorly to the utility that H.264 yields across current telco/broadcast deployments.

Thanks for the compliments on the service. Yes, the focus was on the video engineering so we’re in need of some CSS love.

I do wonder what guided your choice of technology. Most video is published using Adobe Flash these days and it requires only a small server-side extension to provide random access. I would think that rolling out a RTSP base solution is technically more challenging – the more to congratulate you on the success! Were there any other reasons for choosing a QuickTime based solution?

I agree, we are in very interesting times – just like when the first Web content management systems emerged and everyone wanted to get on the Web, but it wasn’t clear what the best system choices would be. Right now, a central system that would relieve everyone of having to solve the hard problems (format choice, encoding setup, video player design, accessibility, statistics capture etc.) indeed seems to make a huge amount of sense.

BTW: A few open source video management systems are starting to emerge with VideoPress, Kaltura CE and Fez/Fedora (the latter more of a general content archiving and publishing system).

I think we will see a lot of innovation in the space in the next few years, not just because of HTML5, but encouraged by it. Our NBN is coming at the right time to help us focus on high quality content that should even be re-usable on IP-based TV with a good experience.

Agreed, the choice of archive video format for the long term needs to be flexible over time, so that you can ingest everything today and have a mapping/transcoding to whatever might be used in the future. It’s an ongoing decision over the life of the archive.

For example, Parliament House has over 30,000 hours of material on tapes going back to 1987. They currently squeeze current material out to today’s broadband viewers, but in 8 years they’ll probably need to be delivering SD (720×576 @25) or more to home desktops.

I have great hopes for HTML5 being able to standardise browser video. Right now, it’s a mess. We’re putting the finishing touches on this ACT Government site that provides searchable access to audiovisual speeches in the assembly chamber.

ACT Legislative Assembly Daily on Demand
http://on-demand.parliament.act.gov.au/

This site uses RTSP for the video delivery mechanism primarily because we need fast, random access to multiple stored videos. It delivers H.264 via the quicktime architecture. It works OK for the current retail broadband scenario. (although criticism gratefully accepted).

However Australia’s future is a high-bandwidth one on the NBN. I think if the NBN multicast architecture is sound, a video archive could deliver the same speed and random access of RTSP. But more likely is the current methods of delivering H.264 in a flash or HTML wrapper via HTTP will overcome the access/seeking deficiencies.

Yes, GOV, EDU and COM all have the same problem. Each have their own tweeks, but agencies/departments/colleges just need to have a single thing they contribute to. All these organisational entities have some degree of ability and material, so the key is to cause them as little pain as possible. In EDU, some unis have their own lecture recording system, some use an external service, some cobble, some have none.

Interesting times!

George

It is amazing how your architecture matches with a lot of the thoughts we have had as well.

We also don’t want to force massive changes on the agencies, but instead allow them new video functionality within their given Web CMS. A central video solution would provide the functionality of video publishing, but could be delivered through either simple embedding/HTML widgets into a given Web CMS, or through plugins written for a Web CMS.

We are also thinking about a central aggregation/syndication place. It could be a clearinghouse for metadata and rights.

As what concerns video format standards: we are indeed in a turbulent time right now. With HTML5 and the new video and audio elements coming out, Ogg format (as is being used in the Open Video example you pasted later) is the most open and freely available format and is now given a chance to succeed on the Web. Right now, H.264 in Adobe Flash is the most widely spread codec format – but it will require payment of royalties from 2011 for Web publishers, so is not open and free.

In my mind, the format in which we publish audio and video on the Web will still change for another decade or two until we will reach a state where we can transfer audio and video at their highest quality and picture-size, compressed almost losslessly. This is the state that JPEG has got into and it’s not really necessary to invent any more image compression formats. So, we just have to deal with the situation that our publishing system will need to be able to change underlying compression formats for the binary data and make sure that the rest of the system (upload, transcode, metadata storage, captioning, audio annotations, etc) will continue to work even if the file format changes.

As for standardizing on an archiving format: I believe you have to archive the source material in the highest possible quality, so you can re-create compressed publishing format from them at any stage. Fortunately capturing formats are digital format these days, so they should be simple to archive and also simple to transcode without or with little loss of information to another digital capturing format. If possible, we want uncompressed source material. Given today’s storage sizes, an archive should be able to cope with such material.

It’s great to see educational projects thinking along very similar lines to what a government video system needs – and in fact it will also apply to the needs of large corporates IMO.

Thanks for sharing, George!

What a great effort, I’m really behind it.

My background is setting up large video capture/delivery systems for web use, and I’m currently consulting to the Swiss academic institutions to work out a similar plan for multiple (hundreds) of institutions to share and archive their audiovisual media.

In some ways it’s a similar problem. For Switzerland, the aim of the project is to enable each organisation (universities, CERN, etc) to easily make their material available to the national audience. Like any collection of organisations, some will have a strategy for THEIR org, some will be thinking about accessibility, some will be thinking about longevity. All, however, need a common approach to a shared solution that will live long and prosper. We’re talking here about making sure that 2D video presentation from the Prime Minister is available in 200 years on a 4D TV. http://swisseducast.hefr.ch/

There are standards wars being played out now. Both in media formats and metadata schemas. Now is certainly the time for deciding which to use, and that choice really needs to be made with reference to how long you’re expecting to have the media available, and how (if at all) the national libraries and archives are involved for long term availability/searchability.

In the Swiss project we’re considering different approaches, but the most viable model is to require little or no technical changes to an institution’s systems or procedures. Rather, put the engineering effort into building a large aggregation service that pulls material to a central point or references it in-place. Coming up with a coherent metadata plan for all players is a hard part of this. But in the end, the aim is to allow an organisation/department to easily contribute to the pool of nationally shared material.

The central aggregation model can also work well if it’s designed as a clearinghouse. By requiring that all media available from the aggregator conforms to agreed technical, metadata and downstream usage licenses it becomes an environment where the decision to “release” a new piece of media to the public is easy for everyone. Whether those standards can be agreed to is a different matter!

Some existing efforts in this area are the Steeple Project (Oxford) http://steeple.oucs.ox.ac.uk/, and maybe OpenCast http://opencastproject.org

–
George Bray
Managing Director
Sand Consulting Pty Ltd ABN 34 003 208 231
Landline +61 2 6100 8121 Mobile +61 411 111 606

Great suggestions! Exactly the kinds of things we were thinking of. Can we cite you in the report as a “citizen” expressing what you think would be good content?

Silvia.

Neil

just going through your list of requirements in detail – I think it would indeed be nice if it was possible to automatically translate captions into sign language such as Auslan, rendered by a avatar. However, such approaches have not been very successful and are still firmly based in the research world. You can, e.g. check out some information in this presentation: http://www.docstoc.com/docs/13449335/Is-Video-on-the-Web-for-Sign-Languages_

In summary, I think it’s an impossible requirement to ask of a video solution to provide automated Auslan conversion of captions.