Best Practices Advanced

MetaHumans are among the most demanding assets in Unreal Engine. This lesson covers the essential optimization techniques, rendering strategies, and deployment patterns that make MetaHuman projects viable for real-time applications across platforms.

LOD Management

LOD LevelTriangle CountUse DistanceFeatures
LOD 0~260K0-3mFull detail, strand hair, full face rig
LOD 1~100K3-8mReduced mesh, cards hair, simplified face
LOD 2~30K8-20mLow poly, baked hair texture, basic face
LOD 3~5K20m+Billboard or minimal mesh, no face animation

Nanite and Lumen

MetaHuman body meshes support Nanite for automatic LOD without pop-in. However, faces and groom (hair) assets do not yet support Nanite. Use Lumen for global illumination to get realistic skin lighting without baking - but budget for the additional GPU cost.

Strand Hair vs. Cards: Strand-based hair (groom) looks stunning but costs 2-4ms of GPU time per character. For scenes with multiple MetaHumans, switch to cards-based hair at LOD 1+ for a 60-80% performance improvement.

Memory and Streaming

  • Use texture streaming for MetaHuman skin and clothing textures
  • Load MetaHumans asynchronously to avoid frame hitches during level transitions
  • Share materials across MetaHuman variants using Material Instances with parameter overrides
  • Budget 500 MB - 1 GB VRAM per visible MetaHuman at LOD 0

Cross-Platform Deployment

For console and mobile targets, use the MetaHuman Quality Scalability settings. Create platform-specific LOD configurations and material quality levels. Test extensively on target hardware early in development.

Ready to Go Deeper?

Live instructor-led courses from our partners. Affiliate disclosure.