Procedural Terrain Generation with Perlin Noise in Unity

Procedural Content Generation with Perlin Noise in Unity

Procedural Content Generation (PCG) is a powerful technique used in modern game development to create vast, varied worlds without manual design. From the endless landscapes of Minecraft to the dynamic weather systems in No Man's Sky, PCG leverages mathematical noise functions like Perlin Noise to produce realistic terrain. In this tutorial, you'll learn how to build a procedural terrain generator in Unity using the provided PCG Terrain component, focusing on creating a heightmap with three visually distinct biomes. This guide aligns with the Cs7632 assignment and will help you understand noise scaling, hierarchical node graphs, and biome segmentation.

Understanding Perlin Noise

Perlin Noise, developed by Ken Perlin, is a gradient noise function that produces smooth, natural-looking patterns. Unlike random noise, Perlin Noise has coherent structures, making it ideal for terrain generation. By adjusting the PerlinScalar (frequency) and MaxValue (amplitude), you can control the scale of features. Low frequency creates rolling hills, while high frequency adds fine details like rocks or grass.

Setting Up the PCG Graph

Open the provided Unity project and load the PCGTerrain scene. You'll work entirely in the Editor. The PCG generator uses a hierarchical node graph where each node applies noise processing and combines with its parent. The root node defines the base terrain, and child nodes add or modify features. Use the Add Child PCG Terrain Node button to build your graph.

Configuring Nodes for Biome Generation

For the assignment, you need three distinct biomes. A common approach is to use the ProcessParentType with a TrapezoidFunction to segment the noise into bands. For example, set the root node to generate Perlin Noise with a wide range. Then add three child nodes, each with a TrapezoidFunction that isolates a specific elevation band. Configure each child's CombineType to Multiply so that the biome appears only in its band.

Biome 1: Mountains

Create a child node for mountains. Use a high PerlinScalar for rugged peaks and a MaxValue of 0.8. Set the TrapezoidFunction parameters (outer fade start, inner band start, inner band end, outer fade end) to target high elevations, e.g., [0.6, 0.7, 1.0, 1.0]. This ensures mountains appear only in the top 30% of the heightmap.

Biome 2: Sand Dunes

Add another child for sand dunes. Use a medium PerlinScalar and lower MaxValue (0.4). Set the TrapezoidFunction to target mid elevations, e.g., [0.3, 0.4, 0.6, 0.7]. This creates rolling dunes in the middle band.

Biome 3: Canyons

For canyons, use a child node with a low PerlinScalar and a negative MaxValue (e.g., -0.5) to carve valleys. Set the TrapezoidFunction to target low elevations, e.g., [0.0, 0.0, 0.3, 0.4]. Combine with Subtract to lower the terrain.

Using Mapping Curves for Smooth Transitions

For smoother biome transitions, consider using the GenNoiseCurve or ProcessParentCurve. These allow you to define custom remapping functions. For example, a sine curve can create gentle blending between biomes, avoiding harsh edges. This is computationally more expensive but yields visually appealing results.

Saving and Validating Your Terrain

Once your graph is configured, click Save to ScriptableObject at the root node. This serializes your graph to a PCGTerrainData.asset file. Remember to save your Unity scene as well. Use the Validate button to check for errors. If valid, you'll see "Valid!" in the console.

Taking Screenshots

To capture your terrain, press Play and move around using WASD and mouse-look. Use mouse1 and mouse2 to adjust elevation. Take at least four screenshots showcasing each biome and the overall terrain. You can also share them on Piazza with classmates.

Trend Connection: AI and Procedural Generation

Procedural generation isn't just for games—it's used in AI training environments and simulation. For example, the Unity ML-Agents framework uses procedurally generated environments to train reinforcement learning agents. By mastering Perlin Noise and PCG graphs, you're learning skills applicable to AI, virtual worlds, and even financial modeling where noise and trends are analyzed.

Conclusion

By following this tutorial, you've built a procedural terrain generator that creates three distinct biomes using Perlin Noise and hierarchical nodes. Experiment with different noise parameters, curves, and combine types to create unique landscapes. This foundation will serve you well in game development, simulation, and beyond.