9 comments

[ 5.8 ms ] story [ 30.7 ms ] thread
A few steps away from a randomly sparse pixel display you can just string up and display things bounded by density.
Unfortunately, even a 64 x 64 grid uses a surprisingly large amount of power and you'll find yourself working with high current DC power supplies.
Yes, you need to start injecting power along the strings of LEDs to keep it working properly and the power demands do add up pretty quickly. Fortunately it's low voltage (5v) so not shock worthy, but definitely capable of starting fires!
I built a 64x64 grid of neopixels. It used 2-3 large PC power supplies, providing about 1.2kW at peak (all pixels fully on).
I had spent a bunch of time trying to do the exact same thing with self powered esp8266s (one rgb led per device), but never finished. This captures the essence of the coolest part which “learns” how to use randomly placed pixels as a display. I do think there’s a faster algorithm for the calibration step though...
I think you are right, one pixel at a time is a pretty slow way of doing things. The current image processing is pretty naive at the moment as well, gets easily confused by large reflections and splashes of light.
Since LED can also work as light detector, I wonder if it would be possible to calculate a similar positioning map by detecting the flashes of nearby LEDs.
In theory, if you had individual leads to each LED, possible. However, I don't think the way WS2811s (and similar) are addressed would allow this.