Aug 30

Print this Post

Alternative capture agent hardware

When researching audio and video equipment for a capture agent, it didn’t take long for someone to ask “Isn’t there a better way than that?” A lot of auxiliary equipment is needed just to get typical audio and video devices to work with a computer. For example a video capture card is needed to interface with a camera or an audio mixer is needed to power a boundary microphone. Settings like volume  must be adjusted physically on the devices, when the computer shuts down the devices don’t, extra power supplies and mounts are required for installation, etc. In a studio or production environment it makes sense to have this kind of equipment but for installations that will rarely have physical user interaction it doesn’t make a lot of sense. Our intentions are to start with a clean slate and simplify everything related to the audio and video equipment for a capture agent.

Webcams have left a lot to desire in years past. Poor color balance, low resolution images, and noisy images are common problems that come to mind when thinking about webcams. Recently there’s been considerable advances in the quality of images coming from webcams. There are several webcams on the market right now that can capture native HD video without software scaling of the source. These sound like great candidates for a capture agent. They are very simple because no external power is required, there are no settings or adjustments on the camera, it powers off when not in use or when the computer is off, and if the camera is upgraded there are no auxiliary devices that need upgrading too (like capture cards and power supplies).

On the audio side of things we’ve found boundary microphones to provide the best audio capture for an entire classroom. This type of microphone requires external power which traditionally comes in the form phantom power which is power provided by an audio mixer on the same cable as the signal (XLR cable). Recently manufacturers have adopted USB as an interface for these microphones and advertised them as web conference microphones. Instead of getting power for the microphone from phantom power on a mixer it comes from the USB power of the computer (brilliant!). The audio signal can be sent over USB as well which effectively eliminates the need for a mixer in simple setups.

If devices can’t be placed far enough away from the capture agent then it greatly limits how the devices can actually be used and how the capture agent can be installed. This is a major concern that could make or break the deal of using USB devices for a capture agent. Common audio and video cables are designed to go pretty long distances, for example there are XLR cables for microphones as long as 250 feet. We’ve found active repeater cables work reliably up to three cables chained together (or four if you put a powered hub somewhere in the middle).  At 16 feet per cable that gives a range of 48-64 feet plus the length of the device cable which should be enough to meet the needs for a typical classroom. For very large or unusual classrooms there are Ethernet extenders that can go hundreds of feet, but we haven’t tested those yet. The hardware we decided to test for our first generation capture agent is listed below along with the prices paid in August of 2010 in US dollars.

So, what’s the verdict of the initial testing? Frame rates of the video are not as high as we would like them (10 FPS at 720P). The bandwidth of USB 2.0 is insufficient for higher resolution video combined with higher frame rates (you get one or the other), regardless we think there’s a lot of potential there. With USB 3.0 on the market providing ten times the bandwidth we’re hoping that will enable cameras with higher resolution and frame rates (e.g. 1080P at 30FPS). Note the linked Point Grey camera is a concept device, not an actual production model. The USB microphone performs wonderfully, maybe too good. The instructor can be heard clearly but so can papers rustling and doors opening and closing. The camera has a wide angle of view which is suited well for shorter distances. At 720P resolution there seems to be ample detail in the video. Given the sensitivity of the microphone to noise from the environment and the wide angle of the camera, this hardware is best suited for smaller classrooms. Fortunately most of the classrooms here are smaller and more intimate. Having said that, USB devices could be used for larger classrooms but it would be on a case by case basis depending on acoustics of the room and where devices could be mounted.

There are other hardware options like wireless microphones for instructors to reduce noise from the environment but we want the capture agent setup to be as simple as possible and unobtrusive. It would really stink for an entire lecture to be lost if the instructor forgot to turn on the wireless microphone, or if the battery unknowingly died in the middle of class. The capture agent itself is not discussed in this yet because a laptop has been used for the initial testing. Our next step is to build a capture agent specifically for use with USB devices that will be installed in a classroom. Below is a sample video from our capture agent testing before it gets ingested by Matterhorn.

The photo of the projector shows a potential mounting position for a camera. Since there’s conduit already there for the projector cables it can be used again for the camera cables. In classrooms without a projector the camera can be mounted using strut channel or a number of other methods. The last photo shows an actual frame from the captured video at 720P.

Permanent link to this article: http://sites.tamuc.edu/innovations/alternative-capture-agent-hardware/


Skip to comment form

  1. Christopher Brooks

    Looks dope, especially the mic since batteries are the bane of our existence here. What did you use for the actual capture software – matterhorn, or something else?

  2. Lovemore Nalube

    Hi there!

    How did you mae your capture agent work with your logitech c910 USB webcam? I have v1.0 running and it is not picking up the webcam (see: http://opencast.3480289.n2.nabble.com/Pipeline-start-error-on-Capture-Agent-td5647493.html) The word from the community is that it is not supported. This ticket (http://opencast.jira.com/browse/MH-5381) doesn’t have info on how to patch the pipeline code.

    Please help.

  3. Jimmy Gray "The Forex" Man

    I was just browsing for relevant blog posts for my project research and I happened to stumble upon yours. Thanks for the excellent information!

  4. Milan Gruner

    Hi guys,
    I’ve got the same problem as Lovemore Nalube. My camera even isn’t recognized by the system (running on Ubuntu).

    Thx 4 your help,
    Le Milonkh

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>