← Back to the studio

Seed

Lab / 001 / Internal experiment

Same seed. Same configuration.
Mobile uses fewer pieces.

A working experiment in controlled variation

The live generator needs JavaScript and WebGL. The method is readable below.

Variation is not
the same as chance.

Nothing rolls freely

The seed chooses between authored options rather than sampling a range. The few values that are loose have ranges tight enough that both ends are good.

Focal points are placed

Accent count is guaranteed and spread across depth, so no seed produces a version with nothing to look at.

Dependent values derive

Scale and placement follow a shared configuration. Each piece belongs to the same structure, even as the structure changes.

Determinism is absolute

One stream consumed in a fixed order. A shared seed reproduces the same configuration at the same viewport class. Mobile uses a smaller pool of pieces.

What the machine
does not decide.

The timing, the cut, the copy, and the order things happen in. Knowing what to withhold from a generator is the whole craft — a system that decides everything decides nothing well.