Domain-warping noise with noise — feeding noise into modulation-warp — produces turbulent, liquid, marbled texture and is the richest cheap texture

Domain warping (warped noise) means using one noise evaluation to perturb the sample coordinates before evaluating another noise layer. The output coordinates p’ = p + noise(p) are fed back into a second noise(p’) evaluation. Because the displacement is itself smoothly varying (coherent), the result has turbulent, curling, marbled character that plain noise cannot produce. The phrase ‘richest cheap texture’ means the technique has the highest visual complexity-to-code ratio: it requires only one extra step (a coordinate displacement) yet produces results that resemble natural materials closely enough to eliminate the need for much more complex rendering approaches. In Hydra, this corresponds to .modulate(noise(...), amount).

Examples

Hydra: noise(3,0.1).modulate(noise(2,0.15),0.4) — the second noise displaces sample coords of the first, creating turbulent liquid/marble. Increasing the modulation amount from 0.1 to 0.8 escalates from gentle warp to full turbulence.

Assessment

What is domain warping? Draw the signal flow for warped noise (in words), explain why the result looks turbulent rather than just noisier, and state one practical risk of pushing the warp amount too high.