Ever since the first news of Leap Motion's coming-soon gesture interface peripheral became public, I've been particularly curious to find out the answers to two architecture questions:
- How much processing is done locally on the device, versus in the USB-tethered computer, and
- What 3-D sensing scheme is the device based on: a stereo sensor array, structured light technology, a time-of-flight approach, or something completely different?
Now, thanks to a just-published teardown on the SparkFun Electronics website (with thanks to Gizmodo for the heads-up), we have at least some answers. The local processing comprises a Cypress Semiconductor USB 3.0 PHY IC, supplemented by a Macronix 32 Mbit serial NOR flash memory, suggesting that the bulk of the system's intelligence resides in the connected PC. And the Leap Motion peripheral implements a stereoscopic vision arrangement, with three infrared LEDs providing active illumination to supplement any available ambient lighting. At this time, the image sensors' manufacturer and other specifics remain unknown
For more, check out this demonstration video that SparkFun shot, using a visualization application included in the Leap Motion SDK.
