Break Point

Break Point splits a point into its axes, and it has two faces decided by what arrives on the Point input.

• ONE point — a SIMD vector (Make Point, Get Point), an [x, y, z] tuple, or an {x, y} dictionary (a Mouse position) — X, Y and Z come out as plain numbers and Count is 1.
• A whole cloud — any array of those, even a single-element one — X, Y and Z become PER-AXIS ARRAYS: every point's X in one list, every Y in another, every Z in a third. The three pins switch to array shape live, and Count is the cloud size.

That second face is the door into the Array toolbox for point clouds: Map, Sort, Filter, Interpolate, Accumulate, Stats, Reverse and the rest all work on one axis at a time.

As Array forces the array face regardless of the input: an empty or unconnected Point gives empty arrays (Count 0) and a single point gives one-element arrays, with the pins staying square. Turn it on when the Array patch downstream must keep its array typing before the cloud cable is live — otherwise the node reads as "a point at the origin" and the pins are scalar until real data arrives.

• Per-axis statistics — Break Point → Array Stats on Y for the cloud's height range, mean and spread.
• Axis filtering — drop everything below a line with Array Filter, then rebuild with Make Point (Match Indexes).
• One point, three numbers — read a fingertip's X and Y straight into other parameters.
• Reordering — Array Sort one axis and zip it back for a combed cloud.

• A one-element array is still a CLOUD, not a point — the array face is chosen by shape, not by size.
• Cables already attached are not re-gated when the face flips; a scalar consumer fed an array simply reads its default.
• Get by Index and Get by Name also open a single point (x / y / z) if you only need one component.