Image Acquisition

For a successful video conference there are four key components * as far as getting a good picture to the far participants. 

Here they are in sort of a start to finish order:

  • Image Aquisition
  • Compression then decompression after transmission (the codec)
  • Network (transmission)
  • Display

In this blog, I'll discuss the image acquisition portion of video conferencing, as I've found that of the four, it lacks the most attention.

Codec vendors have made great strides over years, not only in quality of the product, but in price. No longer are you forced into spending $20,000 (or more) per endpoint just for the codec. Of course, you can spend that much but it’s no longer a requirement. Soft codecs are also getting better all the time and can no longer be discounted as an inferior product.

Network speeds continue to get faster, service is getting better and the price is such that broadband is easily affordable.

Display technology has boomed and it’s no longer lagging behind the codec in capability like it was a decade ago. In fact, display resolution has surpassed the ability of the codec.

So what does that leave us with? Image acquisition. As an industry, we’re still by and large using the PTZ as our go to camera. In fact, Sony, who makes great cameras just released another one at Infocomm. While their resolution has increased and some features have been added, we’re still talking a large device placed inappropriately on top of the display (at best).  Even in the immersive telepresence systems that exist, only a couple have ventured beyond tossing a camera or three on top of a display. So the question becomes why hasn’t image acquisition become more of a priority? Not only that, but can something be done with the image to improve it before sending it to the codec?

At Array Telepresence, we’ve taken the opportunity to address the problem of both image acquisition and improvement. First, we take a strict stance on where our small camera goes. While in theory it can be placed wherever you want it, we provide a guideline on where to put it to optimize the experience. Cameras for video conferencing should be at eye level to the local participants, provided the displays are hung in a normal location. If the displays are mounted higher, the camera should go with it so they are at the eye level of the participants on screen. This provides the best sense of eye contact which is one of the key problems with the industry.

Cameras should never be placed in locations where there isn't even a semblance of eye contact when looking at the person you're talking to. That's lead to a bunch of my peers writing (poor) blogs insisting that you need to "look at the camera". Read more about my thoughts on bad video conferencing advice here.

Image Improvement
Most of us who have been in a PTZ video conference knows how it goes. People in the back of the room are really small or there is some jerk who won't lay off the remote and keeps zooming in on everyone. Worse yet are "auto PTZ" cameras that try and track people who are talking. These cameras simply aren't fast enough to keep up with different people talking over each other in a room, you know, like what happens in a real meeting. Remember, video conferences are only supposed to be meetings with people who aren't in the same room. That's it. That's what needs to be replicated as best as possible.

We've solved the problem of tiny heads in the back of the room. We have developed a piece of hardware that sits in front of your existing codec that equalizes all the participants in the room. No more postage stamp sized heads at the back of the table. No more panning, tilting and zooming as our camera is fixed capture. At that point, there's really no need for the $300,000 immersive telepresence systems either (which was originated on fixed capture and life sized participants across the entire table), unless you like spending money.

*In addition to image acquisition, there is also audio capture and transmission which is equally if not more important. But this blog isn't about audio and therefore didn't write about it. 

If you have any questions about Array, the product or my involvement in its development let me know.

Follow me on Twitter @bryanhellard

No comments:

Post a Comment