TypologySienna Chen
InputLightweight

Procedural seed

One number decides all the randomness.

One number that decides all the randomness, so the same number always gives the same artwork. It makes generative pieces shareable by link and reproducible. One stray random call breaks it.

Breaks when
Math.random() sneaks in and breaks reproducibility
Also called
seeded random, fxhash, hash-based

Prompts

Seeded generative piece with permalinks
A generative artwork whose every random value comes from a seeded PRNG (mulberry32 or sfc32) initialized from the URL hash; no Math.random() anywhere (lint for it). Controls: next / previous seed, a random seed, and copy link. Render at 1080×1350 with a Save PNG button that names the file by seed. Show that the same seed renders identically after reload.
Seeded noise and shuffles
Build the seeded toolkit: a PRNG, seeded simplex noise (permutation table from the PRNG), seeded shuffle, seeded pick-weighted, and a hash for deriving sub-seeds per element (so adding an element never changes the others). Unit-test determinism across 1,000 seeds.
fxhash-style features
Derive named features from the seed (palette, density, symmetry) with fixed probabilities, expose them in a small panel, and make the artwork depend only on those features plus the seed. Explain why features must be computed before any drawing.

Related

Resources · 1