#lightbikes

#lights #luces #lightbikes

And if Blake is riding behind her like Olivia Wilde, all the better.

It's Tron Tuesday, so my followers know what that means.... TRON: LEGACY CONCEPT ART!!!! All these are from the Tron Legacy Second Screen App for iPad. I don't own these, Disney does. Simple as that.

This video is by no means easy to understand but here's a quick rundown. The video content inside the yellow outlined shape is what gets sent to the motion tracking processing algorithm. Everything that is black is not really processed. I learned that when it comes to computer vision, it's best to process only the bare minimum of visual data necessary for the task at hand, hence the yellow sliver. 0:00 - 0:33 show the motion tracking working decently: the rider passes through the sliver and a lane (right most portion of the window) lights up (turns from gray to white). At this point I hadn't made the connection from the rider's position on the track to which lane would light up, which is why the same top two lanes light up regardless of the rider's position. 0:34 - 0:37 is the reason I had to abandon the motion tracking method (especially 0:36). Because of the fact that the camera is off to the side of the track, perspective becomes an issue. In other words, when a rider on the inside of the track passes the camera, his/her body takes up the entirety of the sliver, therefore making accurate tracking virtually impossible. As if this wasn't bad enough, the proximity of the rider's body to the camera has the effect of reflecting all of the infrared light regardless of whether the rider is wearing reflective material or not, which would erroneously trigger lanes. As I confirmed by talking with Kyle McDonald, the creator of the ofxCv library for OpenFrameworks (the code that performs the motion tracking), the only way to alleviate this problem would be to mount the camera on a 12ft pole so as to obliterate the problems of perspective. This solution was simply not in the cards. In addition to this problem, 0:41-0:46 shows that occasionally noise would trigger the motion tracker and light up lanes without a rider present. So instead I resolved to doing this interactive light installation the old fashion way--by hand! As the riders rode by I would hit keys 1-6 on my keyboard to trigger the appropriate lane.

3 days till race day...

red hook crit site test II; 9 days till race day...

Software update: got rid of flickering lanes and changing colors

first site test, 2.5 wks before the event