Procedural landscapes and fractal noise
Procedural landscapes often sum noise at multiple frequencies. Broad octaves shape continents and valleys; finer octaves add ridges and surface texture.
Frequency, amplitude and persistence
Each octave increases frequency and usually reduces amplitude. Their ratio controls roughness and approximates a self-affine field over the represented range. Band limiting is essential so frequencies smaller than a pixel do not alias.
Multiple octaves add coarse and fine structure. Persistence controls amplitude decay, while lacunarity controls frequency growth. Continuing above the Nyquist limit creates shimmer and moiré rather than detail. Render-aware band limiting or analytically filtered noise keeps the same landscape consistent between preview and high-resolution output.
Break the obvious grid
Domain rotation and warping prevent axes and repeated octave structure from becoming visible. Ridges, terraces and masks reshape the scalar field into distinct materials and landforms, while deterministic seeds preserve reproducibility.
Rotating octaves relative to one another and warping their lookup coordinates removes visible axes and periodic repetition. Excessive domain warping produces rubbery loops. Ridge functions, terraces and masks should have deliberate roles for rock, channel or sediment. The seed preserves reproducibility and should not reroll silently when another parameter changes.
Let processes edit the noise
Hydraulic and thermal erosion organize random height into drainage and talus. Vegetation, snow and sediment respond to slope, height and moisture. The most convincing procedural world uses fractal noise as geological raw material, not the finished explanation.
Hydraulic erosion couples rainfall, flow, sediment pickup and deposition; thermal erosion limits overly steep slopes. Even simplified steps create valleys, fans and talus causally related to the initial noise field. Vegetation and snow can then respond to elevation, moisture and slope. The result reads as landscape rather than merely a lit texture.
Sources and further reading
This article summarizes the following specialist sources in original wording. Accessed and editorially reviewed 12 August 2026.
Continue the idea
Mountains, surfaces and roughness
How self-affine models describe terrain roughness, why vertical and horizontal scales differ and what erosion adds.
Clouds, turbulence and scale
Why cloud boundaries and turbulent fields show statistical scaling rather than exact geometric self-similarity.
Rendering images larger than one texture
How tiled, disk-backed rendering creates enormous images while preserving coordinates, effects and antialiasing across seams.