Following on from my minor rant about Encoder Irritations, I've been having a quick stab at getting Compressor installed on our processing cluster. Quick though, is not the correct word.

First up, we had to deal with the issue of the OS believing it did not have a graphics card equipped. This took a (first) trip to the machine room to attach the rack monitor to the relevant machine. Now I come back to find that I can't access the system by Apple Remote Desktop (VNC by another name) - all I get is a connection and a black screen. I know from the monitor and the ARD status page that the system is on, working and displaying the login screen, but all I see is black.

Attempt to restart the ARD service on the server, no effect, still black upon connection.

So, to prove a point, I take another trip back to the machine room and disconnect the monitor. Return to office and connect, successfully with ARD, and can now see the server's screen and interact. Third trip to the machine room and I reconnect the monitor, see the desktop (as I had logged in before I left the office) and return to find ARD is still working. At least now I've got access, and there's a monitor attached to the server, let's try the Compressor install.

Attempting to install from our Final Cut Studio 2, and we at stopped first with the error that the resolution is too low, and that "Final Cut Studio Installer requires that your system have a Quartz Extreme capable video card". One change of the resolution and a restart of the installer, and now I'm faced with just the Quartz Extreme message.

This is a latest generation XServe, with an NVidia GeForce GT 120 PCIe graphics card, complete with 256Mb of VRAM (twice the required amount requested by the installer previously). The next image shows you the System Profiler information for the server, just for reference, taken with the monitor attached and accessed via ARD. The line that says "Status: No Display Connected" puzzles me somewhat.

Going through Google and Apple's Support pages reveals nothing of help.

• http://support.apple.com/kb/HT2863 - tells me that the requirements for Quartz Extreme are pretty low in terms of modern graphics cards. It links to... 
• http://support.apple.com/kb/TA22294 - which gives me two links; the first of which is broken and should describe Quartz Extreme's feature, the second is to the developer site with code samples. 
• http://docs.info.apple.com/article.html?path=Mac/10.6/en/8749.html - tells me about 10.6 Help and that the above System Profiler image should have a value on display for Quartz Extreme. I can't see it, can you? 
• http://support.apple.com/kb/SP543 - tells me about Technical Specifications for Compressor 3, all of which I'm exceeding handily. 
• http://support.apple.com/kb/TS1888 - tells me about Compressor Troubleshooting, but only applies once it has been installed. If offers instructions on removing Compressor, but the reinstall instructions state: "From your install disc, reinstall Compressor and Apple Qmaster (and Motion, if applicable)."

So, how did I solve all of this and actually get Compressor (and QMaster) installed?

I rebooted the server.

After a discussion with a colleague, we were reminded about a similar problem on another machine recently where the OS was mis-reporting the graphics capability and resolutions on offer, but was resolved after the hardware (a projector in this case) had been left connected and active whilst the machine was rebooted. Our current theory is that there is a startup process that checks for attached display hardware, and if it does not find any, does not start the graphics processes, such as Quartz Extreme. Unfortunately, we don't have any further information to back this up, nor do we have much time to investigate further.

Based on this experience, it seems to install compressor in the rest of the cluster I will have to:

• Attach the monitor to the relevant system 
• Reboot the server 
• Then install Compressor

Rebooting servers is not a trivial operation, and not one to be undertaken lightly. Having suitable hardware in place, but being stymied by a dumb installer (is Compressor even going to work when there is no monitor attached?? If it is, then why have one attached in the first place??) that is held up by what seems like an OS bug, is not going to win friends.

And all this before I possibly come to the conclusion that it isn't any better than what we've tried previously, just different drawbacks...

Carl

Somewhat of a placeholder topic, but I'm thinking a little bit of investigation could be useful into how podcast/media content from universities could be fed out to home entertainment/multimedia systems (for example XBMC) and if easy, perhaps further work for someone to investigate how worthwhile this could be.

Given the rise of computers/DVRs/games-consoles in the home environment, and their increased connectivity and desire to access content, this could be an interesting (and not necessarily tech oriented) audience to address, perhaps on the grounds that they are naturally inclined towards online content.

Carl

P.S. Yes, it's been a day of sorting through emails and reports, hence the outpouring of blog notes ;-)

In a conversation about how to capture activities in a teaching and learning space, it was highlighted that not all teachers/presenters use the technology, and that some, for example, need to generate content as they go (think whiteboards and flipcharts). This came to mind again last week when Peter and myself were pulling together ideas for the presentation of the Steeple findings and wanting to be able to communicate this with our project partners. Given we were working in one of our flexible teaching/learning spaces, the following image might give you pause to consider how do you go about capturing the activity, and more importantly, automating that process.

The jist of the conversation previously revolved around how to be flexible enough to capture all relevant activities and information for remote viewing, and that, after much discussion about pros and cons, the answer was that it would need a human operated video camera approach. This has a range of pros and cons, but I think mostly cons.

Either way, more questions than answers abound so far, for example: If the activity was pedagogically sound, what technological approach could be taken to make it effective? And if the activity isn't perhaps as well executed for remote understanding, but the need to capture and distribute was still there, what compromises are worth making?

Due to this being on the fringe of pure Steeple activities, I can't devote much time to pursuing this, but I can appreciate that this and the resultant outputs can have an impact on the activities that can be supported by the technologies and processes we're working on.

Carl

There's been some progress with our Capture Pilot since I last wrote, so here's a wee summary so far.

Of the 4 rooms within the pilot:

Evenlode - is now fully wired for capture, supporting: audio inputs from a wireless microphone, two in place Windows PCs and from an auxiliary (laptop) input; video capture of the VGA outputs of either Windows PC or the auxiliary input (laptop or visualiser, etc), the specific input of which is user selectable via a hardware signal matrix on the desk (the same that is used to choose the outputs for the two room projectors). The ceiling mounted camera is wired to the Canopus ADVC, but not connected to the Mac Mini. The Mac Mini is now configured as a 10.6.2 system and enabled for remote management, and accepts a mixed audio input from the room's audio mixer, and video via the Epiphan VGA2USB device. The Mini isn't entirely headless now, I've connected its output to a monitor on the desk (shared with one of the presenter's PCs) and there is a keyboard and mouse locked away for local use if I need to fiddle directly.

I have since tested and proved that it is not possible to combine the VGA2USB device with the Canopus ADVC device into the same Mac Mini. Plugging in the ADVC via the Firewire port appears to cause the default video device (the Epiphan) to be lost and made inaccessible. The lack of a preference pane in OS 10.6 to allow selection of video devices similar to how audio devices are controlled is irritating. Quicktime X can not at present recognise the VGA2USB device (I believe this is due an issue with the Epiphan driver) and thus alternative capture to a local file via QtX isn't possible. The Podcast Capture utility allows the machine to bind a single video and audio device combination to the Podcast Producer server, and allows you to select the video device in the preferences, but this only seems to affect the Podcast Capture application and the conflict with the ADVC seems to make this a moot point presently.

Initial tests show that we can trigger capture remotely via the Podcast Producer web application, and via the local Podcast Capture application if I interact with the Mini directly. The VGA2USB device appears to handle input switching (via the VGA matrix box) in its stride, even with changes to display resolution. More testing is needed on this, especially with regard to inputs that try to output in widescreen format, rather than the 4:3 aspect of the projectors. It is hoped that next term the pilot will advance from being an experimental setup to being used by a selected number of presenters allowing us to assess the usability of the setup and the quality of the outputs. However, this is on hold whilst the main Podcasting Cluster is still without a populated Open Directory.

Isis - is proving to be a more difficult room to configure. Whilst I had had earlier testing of capture via either the ADVC or the VGA2USB, the failures to work there have at least been confirmed on independent hardware (i.e. Evenlode, as noted above). However, the wiring in Isis is still our main problem. Whilst the Mini is configured as in Evenlode, and the video switching and capture is all fine, the audio capture is currently problematic. The APart audio mixer is either exhibiting a fault or an undocumented limitation, in that of the two independent output zones, only one contains both the microphone signals and the stereo mixes, whilst the other has only the stereo mixes (i.e. the sound from the PCs, DVD player, laptops, etc) but no Microphone signals. Consequently, capturing someone talking in there isn't possible at this time. We're in conversation with our external AV consultant as to possible solutions, but with a negligible budget that can't support any change that impacts on the AMX control system (which most obvious solutions require), our working around this issue is ongoing.

Cherwell - is our Apple Certified Training Centre and consequently full of Apple hardware. The ability to capture locally in this environment is all functional, however as these machines are managed separately from the headless Mac Mini's used in the other teaching rooms, binding of key machines to the capture cluster has not yet been done. Part of the holdup is addressing how this can be implemented in accordance with the imaging and machine management policies in place for the centre.

Windrush - has a configured Mac Mini and Epiphan video capture hardware in place, but has not yet been setup for audio capture. The room itself, being so small, doesn't contain a wireless microphone (the hearing aide systems are all wired completely independently from the the room audio setup) and has no other microphones currently linked to the room audio system. It is further complicated by having a podium setup that contains two monitors, wireless keyboards and mice, a VGA matrix and Auxillary input wiring, but which is separate from the rack containing the two Windows PCs linked to the podium monitors and matrix, the audio mixer and amplifier. This poses some interesting wiring concerns which are being discussed with the room's stakeholders. The Mini and capture hardware can fit nicely on the podium, thus are close to the source of the VGA signal, but then have no access to the audio from the PCs. Voice capture may be addressed by introducing a Samson IS Track USB microphone to the podium, but that has not yet been trialled, and a quick test has raised doubts about the quality of the captured audio from a statically placed microphone. On the plus side, the Samson device also has a stereo mini-jack input for external instruments, so it is conceivable that a small audio cable could be routed between the rack based mixer, and the podium based Samson, thus reducing the amount of wiring needed to be installed. Time however has not been available yet to explore these options further.

... so, did I mention this was a wee summary?

As you can see, there are still a number of minor issues that need to be addressed before an effective trial of this setup can be done, but as is the way, these problems are being addressed as time and resources allow. At our recent Benefits Realisation assembly I was able to have a chat with some of our community partners and learn more about their capture setups, including ELTAC's commercial setup and a few others, many of whom are in a similar position of trying to patch together solutions on tight budgets. If you'd like to share your experiences and setups, please leave some comments below.

Carl.

Along the way we've been working with various different encoding engines and applications, and so far none have done exactly what they claim to do or have limitations that make them impractical for use. I thought I'd share a few comments on some of my recent irritations caused by these would-be-solutions.

Telestream's Episode Pro

Useful in that it can be driven from the command-line, thus lending itself to many forms of batch processing solutions (including Podcast Producer), and in its claims to support a wide range of input and output formats - so many that it should be able to cope with just about anything we throw at it. Unfortunately that's false. Formats that should be supported aren't (too many to list), it is unreliable and will fail without helpful feedback (the desktop application is especially flaky and has continued to be through near 18 months of use and updates), and to do any form of setup requires using a desktop app to generate the XML configuration files - which are then nigh on impossible to edit via any other means than via the desktop application, making fault diagnosis (such as a misconfigured setup) very difficult to do.

While the H264 outputs and range of configurability is pretty good (both in terms of rendered quality and range of functions), the lack of support for multiple processors in the pro product does limit encoding rates, and the poor experiences with the server and desktop products have caused us to stay well away from the expensive Episode Engine offering. That said, by restricting what inputs we will accept to a narrow band of known working formats (to the frustration of our users), we continue to use Episode Pro with Podcast Producer until such time as another solution offers itself, and that brings us on to...

Apple's Final Cut Compressor

With Apple's recent rewriting of Quicktime they have removed many options for changing the configuration of outputs produced by the Quicktime system. To counter this, they have placed an emphasis on the use of their Pro video software and the Compressor video encoding application. Awareness of this application has been made in the Podcast Composer tool that comes with Podcast Producer 2, and in effect, this is the only means to be able to specify outputs that are not compatible with only Apple products - that they have effectively hidden the specification and format of the pre-defined outputs that they do offer now also irritates as we can not trust them to be what we need.

So, this brings us to trying to use Compressor. Trying being the operative word. I've had negative feedback in the past about the limitations offered by Compressor's wider, but restrictive, configuration panels - to give a specific example: Our desired output formats call for a H264 encoded video output with a width of 640 pixels and a height determined by the aspect ratio of the input. That output has a baseline encoding setup and enforced keyframes roughly every other second. It also produces (via Episode at least), acceptable quality output at an average data rate set to 700kbps, resulting in files approaching 400MBs per hour of content. Compressor however, doesn't believe that 700kbps is an allowable option for VGA (i.e. 640 pixels wide) output, and will only let you choose between 1000kbps and 1500kbps, consequently ensuring that an hour's worth of content now needs in excess of 500MBs space.

However, compressor is supposed to be far more efficient for running on our podcasting processing cluster and (I believe) able to slice up single inputs and spread the encoding load over multiple cores for just a single output file (thus, truly parallising the encoding process, rather than just the batch jobs). That with it's integration to Podcast Composer and use by Final Cut Studio and Server (the former of which is used extensively here for content creation, the latter of which is soon to be evaluated for use in the podcasting service) means I need to seriously consider using Compressor, but so far, i'm just finding barriers to its use. Another of which is the rather moronic bug created by Compressor's installer that means I can't even get it installed on our podcasting cluster.

The image above shows Apple shooting themselves in the foot by stopping an installation from happening on XServes in a machine room cluster. The systems are effectively headless, being managed by remote administration, and by nature of them being in a machine room rack, they do not have bulky monitors attached to each unit - indeed, there is but one monitor to share between the several machines in the rack. Instead of accounting for the obvious, there is a bug in OS X (which I can find no indication is being addressed) that reports there to be no graphics hardware in a machine unless it has an active display attached to it. This doesn't appear to be limited to the XServe hardware either, I had a similar report from a Mac-mini that would allow installation with a monitor attached, but gave the same critical stop message when addressed as a headless system.

The upshot of all this is that I'm going to have to spend time in the machine room physically attaching a monitor to each server as I attempt to install Compressor, just so that I can confirm and test the abilities of the application as an encoding solution, with the anticipated result of determining that it is not a complete answer to our processing needs, or perhaps not even suitable for being part of them. FFMpeg is looking like it needs another evaluation in the context of integration with Podcast Producer...

Carl.

P.S. The title should be a clue to understanding the tone of my reporting :(

I'm writing this note whilst being a little behind on the news, and partly to pull together a few bits and pieces that have happened over the past year and some that are upcoming.

Back in January of 2009, a few months prior to my actually joining the Steeple project (though I was on the periphery at the time), I attended the 1st Media Management and Distribution Workshop (http://www.terena.org/activities/media/ws1/) run by Terena (Pan European organisation of higher education networks - http://www.terena.org/) and hosted by Switch (the Swiss NREN, or JANet equivalent).

I put together a report on what was discussed and learnt there (I'll find the link later) and the knowledge and content has come into use several times since (summarised at our Workshop in March, discussed during the Beyond Walls conference in April, touched upon when meeting other attendees at WWDC in June, etc). The talks are also available via video streams (produced using Switch's own SWITCHCast system for media capture and distribution) on the website - you're just lacking the conversations held around the talks during the two days there ;-)

In short, this forum came together to help the various Institution, National and Multinational initiatives related to Media (Podcasting, Streaming, etc) in Education to meet, share experiences and build relationships. The low level of UK attendance (I counted myself and 1 other from the UK IIRC) was a reflection on how far behind our near neighbours we have been in terms of supporting the use of Multimedia capture and distribution within UKHE - I have been quoted before as saying we're 2 years behind Europe and about 3 behind the US.

Anyhow, there was a mailing list for those interested, the archive of which can be seen at http://www.terena.org/mail-archives/vrrd/, and from that starting point Terena have been working to put together a Task Force for Media Management and Distribution or TF-Media for short. That task force is likely to start officially in January 2010, but there is a preparation meeting that is taking place in Prague tomorrow and Friday. Worry not though if you're hearing about this just now and can't get a flight in time (myself included), they are working to broadcast the meeting and allow remote participation.

Details about the event, and a link to the remote participation (via Adobe's Acrobat Connect Pro system) can be found at http://www.terena.org/activities/media/ws2/. I'm hoping to be virtually around for parts of the two days, and my partner in cri.. ahem, Steeple, Bjoern can be found to be actually talking about Steeple activities as part of the two day agenda.

I'll leave you with this block of text which outlines the Task Force's objectives:

Task force objectives

The overall aim of TF-Media is to collect and exchange ideas, knowledge and experiences about the technical, administrative and legal aspects of multimedia production, management and distribution workflows on a pan-European scale.

It also aims to identify, coordinate and where possible, collaborate on national activities related to different rich media recording, editing and streaming systems and services, as well as federated repository service developments and deployments.

The specific objectives are:

• To provide a forum for exchanging and promoting ideas, experience, and knowledge about the technical, administrative and legal issues.
• To identify and promote the use of open standards, architectures, technology and procedures in the field of multimedia production, management and distribution that are relevant to the educational and research community.
• To foster the federation of the national media content and metadata repositories at a pan-European level.
• To identify and coordinate and where possible collaborate on the national activities related to multimedia recording, management, streaming, as well as the repository service development and deployment.
• To facilitate making guideline or best practice documents containing and promoting information about the existing standards, workflows, latest technology and service requirements, as well as rollout programmes and possibly also operation of common services.
• To work together with related TERENA task forces such as TF-Storage, TF-EMC2 and TF-PR.
• To liaise with other communities such as the Opencast Community, EUNIS and standardisation bodies.

Carl.

It is free to use for anyone and will be a great benefit for those who work, visit or live in Oxford. Imagine you're travelling back from a weekend away but you need to prepare for an essay this evening- you can now ask the system to tell you where you are when you're dropped off, where the nearest college library is, if there is a podcast available by your tutor and what time is the next bus home.

The site hosts an array of features including access to real time bus information from across the county, location information about University buildings and general points of interest, including geolocating where you are and telling you the nearest library or restaurant).

The Steeple project was involved with helping deliverthe instituional set of podcasts to this mobile portal. If you visit http://m.ox.ac.uk on a web-capable phone, you will also be able to run an application that provides a nicer interface for listening/watching any of the Oxford podcasts. So, the mobile phone becomes yet another delivery channel to content, fed automatically from the institutional RSS podcasting system.

The software that underpins Mobile Oxford is part of the JISC funded Erewhon project ( a sister project to Steeple) and the Erewhon project outputs will be useful for other institutions.

One aspect that seems to be often overlooked is ensuring consistent presentation of metadata in various locations, notably inside the content file itself. Mostly as a reminder to myself I'm going to use this posting to make some notes on the various file formats supported as outputs and some information related to the extent and format of the metadata those files formats can take.

This is a Work In Progress :-)

MP3

http://www.digitalpreservation.gov/formats/fdd/fdd000106.shtml
Handy information guide leaning towards library level metadata. Also contains links to related specs.

http://en.wikipedia.org/wiki/Comparison_of_audio_codecs
Whilst I'm at it, audio comparison table on Wikipedia

MP4

http://atomicparsley.sourceforge.net/mpeg-4files.html
Atomic Parsley has some useful notes about the metadata elements in the MPEG4 format

http://en.wikipedia.org/wiki/ITunes
Wikipedia notes a few things about iTunes supported file formats and their metadata:

"iTunes 9 can currently read, write and convert between MP3, AIFF, WAV, MPEG-4, AAC and Apple Lossless."

"For MP3 files, iTunes writes tags in ID3v2.2 using UCS-2 encoding by default, but converting them to ID3v2.3 (UCS-2 encoding) and ID3v2.4 (which uses UTF-8 encoding) is possible via its "Advanced" > "Convert ID3 Tags" toolbar menu. If both ID3v2.x and ID3v1.x tags are in a file, iTunes ignores the ID3v1.x tags."

So many different subsets/forms of MPEG 4 listed at: http://www.digitalpreservation.gov/formats/fdd/browse_list.shtml#m

Ogg

As an alternative format to handle the tiny number of use cases not met by the more mainstream, a little effort is needed to look further into this container format for both video and audio applications. The primary source of this information can be found via: http://www.xiph.org/

Xiph state their purpose as:

"The Xiph.Org Foundation is a non-profit corporation dedicated to protecting the foundations of Internet multimedia from control by private interests. Our purpose is to support and develop free, open protocols and software to serve the public, developer and business markets."

Flash

Flash media files are a necessary evil for a small subset of applications, and they too have metadata within them. Some starting points of reference: http://www.devx.com/webdev/Article/31420

WMV

Placeholder.

Peter's talk, a Pocket University, wraps itself around podcasting, iTunes U, mobile portals, second life, open content and much much more.

Carl.