[Matt Pandina] has been documenting his build of a very nice light painting bar on his G+ page. His light painting bar has 64 RGB LEDs being driven by an ATmega328P and four TLC5940 chips. He wrote his own libraries to talk to the TLC5940 as well as his own libraries to pull images off of a MicroSD card. He also wrote a cross-platform program that automatically converts a directory of pngs to something the TLC5940s expect. He says the secret to getting his24-bit color correction looking right is gamma correction. It seems that when the LEDs were run too bright, he couldn’t get the colors quite right. In case you’re curious, those images are 15 inches tall!
You can follow along through his posts as he starts with just a few LEDs and slowly updates and grows it to the impressive state it is at currently.
uh, no you can’t. at least i can’t see it there besides the color cycling leds point.
Very technically nice build, I’ll be keeping an eye on this.
Not coming from a world of digital art, etc, I hadn’t understood the purpose of gamma correction till now! The results look very nice!
Can’t wait to see something like this transition from the breadboard to a more refined product!
I blocked google plus as well as facebook, and I hope we won’t be getting a lot of facebook and google+ links now on HaD, that would not just be annoying to me personally (which would not impress anybody) but would actually be rather sad in general I think.
I am on G+ but I couldn’t really find a lot of details either. G+ is one single most horrendous piece of shit of ever. Everything you post there gets drowned in crap in a matter of hours and there’s no way you could find it later.
The results look pretty good but the color problem is not related to gamma. The first issue with LED is doing proper conversion of RGB into XYZ. You can correct for gamma and do other calibration after that conversion is done.
Here is the math and a 16 bit PWM table for converting 8 bit sRGB that I built for the CYZ_RGB project: http://code.google.com/p/codalyze/source/browse/cyz_rgb/trunk/pwm_logtable.h
Hmm just looked a little deeper; looks like HAD got this wrong in the writeup not the author. He never said that there was any trick — he just mentioned that he was doing gamma correction. He works at Adobe after all; that’s enough to be pretty confident that he’s getting the color math right.