All fields are generated via the Fractionally Integrated Flux (FIF) model as described here, the multifractal cascade method introduced in 1987 by Schertzer and Lovejoy.
A remarkably wide range of scale-invariant fields can be described with three parameters:
- H: smoothness — strength of large-scale fluctuations vs. small-scale ones.
- α: extreme events — lower values produce occasional giant spikes.
- C₁: clustering — when large, activity concentrates in small regions.
Per-axis kernel. Each axis can be acausal (symmetric, the default), causal (one-sided), or odd (antisymmetric, producing a signed field). Causal and odd cannot coexist in the same field — they use incompatible observable kernels.
Relief view renders the 2D field as a line-drawn isometric surface — multifractal spikes become sharp peaks, smooth regions become rolling hills. Drag to rotate; scroll to zoom.
3D mode treats the third axis as time; play through the field as an evolving sequence of 2D slices, with smooth interpolation between source frames.
Listen as sound (1D only) plays the displayed 1D field as raw audio at an 80 kHz sample rate, looping until you click again or change a parameter. Enabled once the grid size is at least 218.
All fields generated using the Python package scaleinvariance, compiled to WebAssembly.