Customizing HLOD Clusters for Visual Fidelity

Adjust clustering parameters to balance the visual accuracy of static meshes against the performance gains of simplified proxies.

In this lesson, you will learn how to fine-tune Hierarchical Level of Detail (HLOD) clustering parameters to strike the optimal balance between high-fidelity visuals and low-cost draw calls in your large-scale environments.

The Balancing Act

When you generate HLODs, Unreal Engine groups static meshes into clusters and merges them into a single proxy mesh. If your clusters are too large, you lose visual nuance and silhouette accuracy; if they are too small, you increase your draw call count, negating the performance benefits of using HLODs in the first place.

Configuring Clustering via HLOD Layers

To customize this, you must navigate to your HLODSetup asset. Inside the Cluster Generation settings, you will find parameters that directly impact your proxy fidelity:

Desired Bounds: This controls the spatial size of your clusters. Smaller values create more, finer-grained clusters, which helps maintain silhouette accuracy for complex shapes but increases the number of actors in the scene.
Simplify Mesh: When using the "Simplify" method, the Max Deviation setting is critical. It determines how far the proxy mesh is allowed to deviate from the original geometry. Lowering this value keeps the mesh shape tighter to the original, though it increases the poly count of the resulting proxy.

Practical Tuning Strategy

Don't rely on global settings for every object type. Instead, create specific HLOD Layers in your content browser. For instance, you might assign a "High Detail" layer to foreground hero assets with a stricter Max Deviation, while using a "Low Detail" layer for dense, repetitive background foliage.

// Example: Concept of adjusting cluster logic via project settings/ini
[HierarchicalLODSettings]
bUseCustomClusterSettings=true
MinClusterSize=5
MaxClusterSize=100
// A higher TargetMeshAccuracy produces higher fidelity at the cost of memory
TargetMeshAccuracy=0.85

By adjusting these values per layer, you ensure that foreground objects remain visually indistinguishable from their source geometry, while distant, densely packed clusters aggressively reduce the vertex budget, keeping your frame rate stable across your World Partition cells.

Key Takeaway: Fine-tuning Max Deviation and Desired Bounds within specific HLOD layers allows you to preserve the visual integrity of hero assets while aggressively simplifying background geometry to maximize rendering performance.