We took the application out into the wild for the first time. Previous beta tests had been conducted in closed quarters. It was time, with the porting of the codebase over to the new commercial Kinect for Windows SDK, to get out and about. What did we learn today with beta in the field? A bunch of things:
- Those minimum hardware requirements are no joke. Testing on a highly portable, albeit somewhat hardware-deficientnetbook (the Asus EEE PC 1201N), was a challenge. Skeletal tracking was the big hog, lagging well behind both color and depth streams.
- Sensor placement is a big deal. Sure, in closed quarters, you can easily predict where things can and should be placed for optimum tracking, but out in the wild you are constantly confronted with the challenge of making the sensor placement work. Is electricity available? Is there a good camera position? Not only that, but while the skeletal tracking is good, sometimes it gets confused. A higher sensor placement may serve the tests better, and we will try it in the future (this time we were ~18″ off the ground).
- The sensor is light. This will be a consideration out in the field. Can it get damaged? Will it withstand any bumps or bruises?
- Encoded files are BIG. Right now, we’re working with video, and video files are large. It’s a common problem, but one that is now at the forefront of our challenges.
- The SDK works with the XBOX sensors. While the liense doesn’t explicitely state that the XBOX sensors won’t work, if you install the SDK on the target client and run your application using an XBOX sensor, applications work. While “…(I) agree that end users of Kinect for Windows Applications are not licensed to use Kinect for Xbox 360 sensors in connection with such Kinect for Windows Applications, and that you and your distributors will not directly or indirectly assist, encourage or enable Kinect for Windows Application end users to do so” (per the commercial license Restricted Use with the Kinect for Xbox 360 sensor clause), it was interesting to test this functionality out.
- People are interested. Curious onlookers were buzzing about what we were doing, which is a great thing.
More public rounds to come, then some online demos thereafter. Good times!