EMFBC

Or how to jump the gun by 6 months, and not check for permission first

Eh What will we aim to provide?

Well, A few of us at London Hackspace have been thinking about how we could make the video side of things better. Much better. We theorize (with past examples in Student TV and Real world TV, that we can reasonably expect to provide:

• HD Live streams for two stages
• Overflow capacity (i.e. "Ahh, stage A is full, can we get the pictures & sound to stage B?")
• All talks, Q&A sessions available on demand within 2 hours of their finish.
• Overnight continuity (i.e. at the end of the day, play out content from previous EMF's, and with permission, content from popular hacky content creators)
• A Cheesy closedown clock, a la BBC style 1am goodnight, hacking forecast, and national anthem.
• Live graphics on the live streams (i.e. Jonty - Beeeeeeees", and perhaps the odd bit of moderated twitter integration).

Isn't this a bit early to start planning?

• No. Finding a good workflow takes time.
• No. Although the cost of making HDTV has dropped massively, we still need to make our own bits (like controllers, not the mixers!!!) to keep costs down: Oddly, the controllers are the most expensive bit on hardware!
• No. We need a while to source everything, and test kit together. Not to mention we'll be running a full test of the Master control room system beforehand.
• No. Much of this planning cannot be done in a multi-threaded fashion.
• No. EMF may not have even picked a site yet, but much of the infrastructure for EMFBC will remain the same. Pretty much the only thing that'll change is cable / fibre lengths, and that doesn't affect us until later in the process.

Can I bikeshed this?

• Not quite yet. Bikeshedding will be available later on, for things like controller design, software for automation, gfx control...
• Chat with us, we'll usually be talking about making TV.
• Yes, if your question is along the lines of "I clean up messy wikis, can I namespace this to my heart's content?"

Topology

Considerations

• Although undecided yet, EMF will likely have 2 or 3 stages, each probably 100 meters or more apart.
• This means we need quite a bit of cable - at these lengths, we have a few issues: at 1.485Gbps (our baud rate) and 100m, we need some pretty hefty coax. This is kind-of-expensive. It begins to be cheaper to go down the fibre route...
• Shorter = cheaper.
• The connection to EMF-NOC will be easy - Cat5e/6 is cheap, and will already be everywhere on the camp.
• HD Encoding and decoding devices will eat up too much space on the network (the cheap ones we could afford effectively packet storm a network, and we'd need a few!)
• Capture (PC), encode > IP> decode (PC) is hilariously unreliable (we need zero failures at all), and introduces far too much latency.
• SO, we are left with only direct connectivity as an option.

Physical topology

• We clearly need to sit our master control room (MCR) next to one of the stages - if our two stages are 100m each side of NOC, we can do a fibre run from MCR at stage A (STA), past NOC (middle), on to stage B (STB). This'll be cheaper than having MCR at NOC, and doing two fibre runs out to STA & STB. To get network, we simply attach at wherever MCR is, and then as EMF will have a lot of bandwidth, we don't need to worry about where we stream content from!
• If we have three stages, we'll pop MCR in the middle one (if there is a middle one), or at NOC if there is no clear shorter way of doing the runs...

MCR or MCV or MCC ?

We need a home. The LHS Bikeshed has already proven that caravans are brilliantly cheap housing for electronics - cheaper than a van, and more prepared than turning up with a pile of stuff. I propose we take this example, and make the Master Control Room a Master Control Caravan. We can then take all the caravan bits out, put our hardware and custom (cheap) controllers in, pre rig everything, and just hook up on site.

The broom Cupboard

An idea by Mr Sands, where we can add interviews, news, and other diversions to our output. As an indicator of the simplicity, we'll need just: 1 camera, a set of floodlights, a few mics, a small audio mixer, and an analog to AES convertor or an embedder (to make the output fit with our master formats of AES3 & HD-SDI. Could this be run collaboratively with EMFM to half the workload?

MCR / Edit / Playout / Storage Network

We're going to get through quite a bit of data over the weekend. Something of the order of around 2 Terabytes. This will be from the captures of the stages, pre-existing content for playout, the edit process, and the final copies (also for playout).

To keep costs down, this can all be on:

• One machine with a decent RAID array. It'll be getting quite hammered, have playout and edit running at the same time. - We'll need something along the lines of RAID10. Basically, redundant, and quick...
• We'll also need an edit machine. Just about Anything quad(or more please!)-core with above 6GB of RAM will do the job. This'll be a dual screen thing, to make it easier.
• An encode machine to compile the edits from the project files will take time. Offloading this onto a machine with as much grunt as possible, will make sure we can keep up with the events on site (a quad core phenom II 9650 will work at 4x slower than realtime, so we really need something nippy to keep up with 2 stages that run for ~ half time on average!)
• One playout machine for each stage : each one will produce a key (opacity, or alpha) and a fill (visible video) to the video mixer. This will allow it to provide graphics, and handle the continuity service. With 2 decklink HD extreme 3d (or similar) units (not decklink duo!) we could consolidate these machines into just one machine. This might be a little hairy though.
• Control PC(s). These will handle connection to the remote mixer(s) from our control panels in MCR, the automation / continuity. They don't need much grunt at all, but will need a few serial ports.