Last week, I wrote about moving in black, where a moving head lighting fixture repositions while it's dark so nobody sees it travel. The catch is that nobody sees it arrive, either. It has to land where the cue says on the numbers alone, so I spent some bench time finding out how good those numbers really are.
Not as good as the spec sheet, it turns out. I taped a mark on the wall about two metres from the head and sent it a full turn of pan. It should've come right back to the tape. It landed 10 to 15 cm to the left, every time, with no wobble, so it isn't gear play: the real pan range is closer to 542 degrees than the 540 the profile says. The second surprise came from an old surveyor's trick. Swing the head half a turn and flip the tilt over the top, and it can hit the same mark from the other side. On this unit the two sides disagree by about 2 degrees, which means the beam leaves the head about a degree off its own axis, and no pan or tilt offset can fix that everywhere. With seven readings, both effects fit within 0.4 degrees, and the fit predicted a reading it had never seen to within a tenth of a degree.
If you're focusing a rig of moving heads, it's worth trying on one unit before the show. Aim at a taped mark, then hit the same mark from the flipped side. If the two don't agree, the light is off its own axis, and your offsets will look right in one spot and drift everywhere else. We're building this into GodotDMX as a short calibration step in the editor, with a check on a mark the fit never saw.