Playing on a 3D system: the rules in short

What a DJ needs to know to make movement in the room work for a crowd. Each rule points to where it comes from: motion perception (the research, part A/B), the MJ playbook (rules checked against the movement lab’s simulation), or club practice. The simulation is a model, not a listening test; the listening test is still to be run.

The one fact

The crowd does not study the motion, it notices it. A move that has to be listened for is a move that did not happen. Everything below follows from that.

The room

  • The crowd is not in the sweet spot. Off centre, transients are pulled to the nearest speaker (the precedence effect): most of the floor hears a moving hi-hat as “which speaker near me gets louder”, not as an angle. Higher-order Ambisonics (A³ uses 3rd order) widens the area where direction holds to almost the whole array. [research A1]

  • Direction errors grow to the sides: about 5° straight ahead, more than 30° at ±90°. Design a bend that must be heard for ≥ 30° as seen from the floor. [research A1, B2]

  • Height is heard worse than azimuth, and only with high frequencies. Fly the big paths low (just above ear height); keep elevation moves for bright material. [research B3; playbook rule 28]

  • Know where front is. Azimuth 0 is front; the performer needs it marked before anyone drags a blob.

  • A visible gesture helps. In a hall, a performer’s visible, synchronous gesture partly makes up for listeners being off centre. [research A1, Marentakis 2008]

What to move

  • The bass holds the floor. Below about 150 Hz nothing localises in a club (room modes, shared subs); practice keeps it steady. Never move anything under ~250 Hz as a path. [research A2; 4DSOUND; L-ISA at the Rex Club]

  • Move the top of the mix. Hats, claps, air, vocal presence: a band of 6–8 kHz at 2 octaves, or 3 kHz at 4 octaves. [playbook rule 6]

  • Never narrower than one octave for a move that should be followed; width matters more than centre frequency. [research A2; playbook rule 7]

  • A narrow high band sits at its own height, whatever the system sends. Don’t try to move it vertically. [research A2, Middlebrooks 1992]

  • With stems, pick the stem that has the top end (hats, vocals) for the clearest paths, and leave the bass stem steady.

How fast

Speeds in laps per bar, because tempo is the unit a DJ thinks in (at 120 BPM, one lap per bar is 180 °/s):

Lap

Reading

1 per 8 bars or slower

a drift, not motion — “the room drifts”

1 per 4 bars

the cruise: a clear path, readable from anywhere in the crowd

1 per 2 bars

energetic; the cruise below about 110 BPM

1 per bar

the ceiling for a path, and only with a wide band above 3 kHz

1 per 2 beats or faster

an effect (swirl, Leslie, tremolo), 1–2 bars at most

~2.5 laps per second

the hearing limit: direction is lost, it becomes texture

  • A speed change must be ×2 to be heard as faster (×½ as slower). Double or halve; never nudge. [research A1, B1]

  • Laps are powers of two in bars, like loop lengths on a deck.

How much at once

  • Two movers; the rest wait. “Less becomes far more” (4DSOUND). Two moving sounds beat one and beat four in every variant of the simulation. In a DJ set usually only two decks play anyway. [playbook rule 13]

  • Never turn several sounds the same way at the same speed — a rotating field can make people feel they are turning (auditory vection). [research A3]

  • At most one audible event per mover every two beats. [playbook rule 15, a hypothesis]

Keep it alive: the breath

  • A steady orbit is heard and then tuned out after a few laps. [playbook rule 19]

  • The breath: every moving sound stands still for beat 4 of every bar and restarts on the one. One whole beat (it must outlast the ~336 ms the ear integrates over), once a bar (twice is a stutter). The most robust result of the movement lab. With two movers, call and response: one breathes on beat 4, the other on beat 2. [playbook rule 18]

  • The beat link is a move on the one, not a wobble on every beat: a per-beat pulse under 20–30° is not heard. [playbook rule 17; research B5]

  • No spatial kicks on the one — they add fatigue faster than interest. [playbook rule 21]

  • Contrast in time beats contrast in space: stops and doublings on bar lines work; jolts and throws are seasoning. [lab log, round 6]

The 3d control

On a stereo channel, 3d decides how much of the channel is in the room:

  • 0 — plain stereo, like any mixer;

  • 0.5 — the filtered band in 3D and the rest steady (the sum is the original track);

  • 1 — only the filtered band, in 3D.

With 3d at zero a moving blob moves nothing you can hear. A first-time user who presses play, watches the blob fly and hears the track sitting still decides the box is broken: “turn up 3d” is step one of any quick start.

Before trusting any of this

The numbers are from the literature and a simulation. The listening test on a real system decides the speed caps, the band defaults and how many movers a floor enjoys.