Flow Field

Evenly spaced streamlines through a noise field — hair, water, wind, made of lines.

What it is

A vector field is built from layered gradient noise; streamlines are traced through it by numerical integration and seeded so that no two lines crowd each other and no gap yawns. The result reads as combed texture — currents, grain, contour-like flow — optionally with stroke width modulated by field strength, or the field curled for vortex structure.

How to use

As texture pages and backgrounds: the even spacing means the page has a uniform tone that colours or shades predictably. Following a single line across the page is its own quiet activity. Pair with a mask (a silhouette left empty) for figure-ground pages.

Purpose

The generative-art staple of the design lane, and its most parametric one: scale, octaves, separation, step and curl each change the fabric of the page rather than its arrangement. The crate owns its noise implementation, so the field is identical on every platform — a determinism requirement, not an optimisation.

History

Streamline visualisation is old fluid dynamics; the even-spacing seeding this crate uses is Jobard & Lefer’s algorithm (“Creating Evenly-Spaced Streamlines of Arbitrary Density”, 1997). Flow fields entered generative art through Processing-era practice (Perlin noise driving particle paths) and became one of its signature textures.

This implementation

• Spec knobs: size, scale, octaves, separation, step, max_steps, modulate_width, curl.
• Generation: own gradient-noise implementation (no external noise crate, for wasm-identical output), RK4 integration, Jobard–Lefer seeding from a queue of candidate points at fixed offsets from existing lines.
• Guarantees: deterministic per seed; line spacing bounded below by separation, which is also what keeps the colourability gate’s ink coverage in range.