Graph-Driven Flow Maps

Overview

I created this flowing structure by simulating the path of particles emitted from vertical lines and guided by Perlin noise fields. Each particle begins at a point between two vertical anchors and then iteratively updates its position by rotating in the direction determined by the noise field. I calculate the angle using the Perlin noise value, convert that into a directional vector, and then move the particle by a small amount along that vector. The paths curve naturally and densely in some regions while dissipating into open space elsewhere, producing smooth arcs and tendrils that evoke fluid motion.

Mathematically, this relies on sampling a 3D noise field with spatial coordinates and a time-like variable to introduce organic variability. Each emitted particle follows a pseudo-random yet deterministic path based on noise gradients, creating patterns that blend structured geometry with stochastic behavior. By layering multiple such lines from randomized base points, I generated a map of currents that resemble airflow or magnetic flux. The final plot uses subtle alpha blending and line weight to emphasize softness and continuity in form.