I assume to populate it, you need an oven? I do IC by day; not PCB. As much as I would love to surface mount my first board, I have never done it.
Angus' lights seem to be doing this null pixel stuff on the IC. I would love to track down that IC. The reason I believe that is, he describes his lights as all having numerical addresses burned in the chip, and the data line routes in parallel on down the string.
If I were into pixels and had a larger yard, the splitters would be a little 0.5" x 1.0" (or so) through-hole board with a PIC12F1501 (8-pins) and differential line receiver (again, 8-pins) installed right next to the pixel string. Each of the splitters could (would) receive the same pixel stream from the head-end.
Right now I've created a PIC code source file in C, with the 'start address' hard-coded at the moment. I'm thinking of several mechanisms for user-provisioning the 'start address' (likely quite difficult and complicated when there are several widgets in parallel, as would be the case in the 'field'). The possibilities are:
1) use jumpers on the future PCB (easiest)
2) the next would be to provide a serial uart input function on an unused pin to allow programming from either a PC, an Arduino, or an 8266 board
3) the same, except over the pixel input pin (almost pointless if there are several of the widgets in parallel).
4) somehow or other decode the incoming pixel stream to obtain the 'start address' (relatively difficult, and almost pointless when there are several of the
widgets in parallel).
Jumpers. Hands down.
In my real-life situations, I am perfectly happy being constrained to exactly 50, 100, 200 null pixels. I'm not so sure that, say, a Boscoyo ornament would feel the same way, though.
Does anybody know if those earlier projects are still accessible?
Barnabybear's design is through-hole and uses an Arduino Nano, a 74HCT126, and assorted resistors. You set the breakout of the splits in the code loaded to the Arduino. Very simple build, I may put one together tonight and test it out.

