I saw this on CNet this evening. Reminds me of Robb’s work with tracking performance artists in 3D space.
This is fun stuff.
It’s wild what you can do in 2D space too. Oh so long ago (97 I think?), Robb inspired me to write a screensaver-like app that continually converted camera detected motion into one of any number of abstract renderings. The coolest used a flocking algorithm. When you walked up to the computer, boids would swarm in a region on the screen that mirrored where you were. Then as you stood still or walked away, the boids would fly away off the screen.
Robb had another version that had 3D ribbons that would rise up and twist and turn in the areas where motion was detected. Then as the motion subsided, the ribbons would fall to the bottom of the screen.
I’ve got to see if any of this code still works.
Robb and I often brainstorm about all the things we’d like to do with real-time analysis of multiple camera images, but generally the computing power is expensive or cumbersome to come by. Robb figures he needs almost a 10,000 times speed improvement in the PC to effortlessly do what he’d really like. I’ll take a 1,000. Yeah, we get a chuckle out of all the people that claim that computers are fast enough. Not for us!
Comments on this entry are closed.