Noise octaves, FBM, turbulence and ridges

Noise octaves, also known as fractal noise or turbulence, represent one of the most powerful techniques in procedural generation. By combining multiple layers of noise at different frequencies and amplitudes, we can create textures that exhibit detail at multiple scales—mimicking the complexity found throughout nature.

Frequency and Amplitude in Noise

Understanding noise octaves begins with two fundamental parameters: frequency and amplitude. Frequency determines how quickly the noise values change across space—high frequency noise has rapid variations with fine detail, while low frequency noise changes slowly with broad, sweeping patterns. Amplitude controls the intensity or magnitude of these variations—how much the noise values deviate from their baseline.

A single octave of noise, regardless of its frequency, tends to look artificial and monotonous. Real-world surfaces exhibit complexity at multiple scales simultaneously: a mountain range has large-scale formations, medium-scale ridges, and fine-scale surface texture. Noise octaves allow us to replicate this multi-scale detail.

Combining Octaves

The standard approach to combining octaves follows a simple but effective pattern: each successive octave doubles in frequency and halves in amplitude. The first octave provides the broad, foundational structure. The second octave, at twice the frequency and half the amplitude, adds medium-scale detail. The third octave adds finer detail still, and so on.

Mathematically, if we start with a base frequency f and amplitude a, the combined noise function becomes:

result = a × noise(f × position) + (a/2) × noise(2f × position) + (a/4) × noise(4f × position) + ...

This summation continues for the desired number of octaves, typically between 4 and 8 for most applications. Each additional octave contributes progressively finer detail while having less overall influence on the final result.

Persistence and Lacunarity

While the doubling-halving pattern works well, two parameters allow fine-tuning of the octave combination: persistence and lacunarity.

Persistence controls how quickly the amplitude decreases with each octave. A persistence of 0.5 gives the standard halving behavior. Higher values (closer to 1.0) make higher octaves more prominent, creating rougher, more chaotic textures with emphasized fine detail. Lower values (closer to 0) make higher octaves less significant, producing smoother results dominated by large-scale features.

Lacunarity determines the frequency multiplier between octaves. A lacunarity of 2.0 gives the standard doubling. Higher values create larger gaps between frequency scales, producing more distinct layers of detail. Lower values create more gradual transitions between scales, resulting in smoother, more continuous textures.

Fractal Self-Similarity

The power of noise octaves lies in their creation of fractal-like self-similarity. When you zoom into a multi-octave noise texture, you continue to see similar patterns of detail at every scale—just as a coastline reveals similar complexity whether viewed from space, from an airplane, or while standing on the shore.

This self-similarity is what makes octave-based noise feel natural. Natural phenomena often exhibit fractal characteristics because the same physical processes operate at multiple scales. Mountains erode through similar mechanisms whether we're looking at kilometer-scale valleys or centimeter-scale gullies. Noise octaves capture this property mathematically.

Applications

Terrain Generation

Terrain generation is perhaps the most iconic application of noise octaves. The first octave establishes major mountain ranges and valleys. Subsequent octaves add hills, ridges, rocky outcrops, and finally surface-level detail like small stones and texture. By adjusting persistence, artists can control whether terrain appears smooth and rolling or jagged and mountainous.

Cloud Textures

Clouds exhibit natural turbulence at multiple scales, making them ideal candidates for octave-based noise. Large-scale octaves define the overall cloud formations and weather patterns, while smaller octaves add the wispy, billowing details that make clouds feel volumetric and alive. Three-dimensional noise octaves can even simulate the internal structure of clouds for realistic volumetric rendering.

Organic Patterns

Beyond terrain and clouds, noise octaves excel at creating organic patterns found throughout nature: wood grain with its large growth rings and fine fiber texture, marble with its sweeping veins and subtle color variations, water surfaces with large waves and small ripples, and even biological textures like skin pores or bark patterns. The multi-scale detail inherent in octave noise makes these materials convincing without requiring photographic reference.

Implementation Considerations

While conceptually simple, implementing noise octaves requires attention to performance. Each additional octave requires another complete noise evaluation, so computational cost scales linearly with octave count. For real-time applications, finding the minimum number of octaves that achieves the desired visual quality is essential.

The choice of underlying noise function also matters. Perlin noise, Simplex noise, and other algorithms each have different characteristics that become amplified when combined in octaves. Perlin noise tends to produce more directional artifacts, while Simplex noise offers more isotropic results. The base noise function should be chosen based on the specific visual requirements of the application.