Explanation
An optimization technique that avoids computing objects that are not visible to the user.
Real-world example
Not rendering the graphics behind a wall to save processing power.
Practical applications
- GPU optimization: saving up to 50% of graphics computation
- Maintaining framerate: ensuring the 90 FPS required in VR
- Complex scenes: enabling highly detailed environments
- Headset battery life: reducing power consumption on standalone headsets
Occlusion culling techniques
Frustum culling
- Eliminates everything outside the field of view
- Simple and fast to compute
- Always active by default
- Insufficient on its own for complex scenes
Example: Not rendering objects behind the camera
Static occlusion culling
- Pre-computed during the scene bake
- Divides the scene into cells and portals
- Very effective for architectural environments
- Does not work for moving objects
Example: In a building, loading only the visible room and its neighbors
Dynamic occlusion culling
- Computed in real time every frame
- Uses the GPU (hierarchical-Z)
- Works with moving objects
- Computation cost must be optimized
Example: Automatically hiding enemies behind a moving vehicle
VR scenario
In a virtual factory tour, the user is looking at one machine. The 200 other machines in the factory are invisible because they are hidden behind walls or out of view. Occlusion culling reduces the triangle count from 50 million to 2 million. Without this optimization, the VR headset could not reach the required FPS and would cause motion sickness.
Why it matters in professional VR
- VR performance: an absolute requirement for maintaining a constant 90 FPS
- Accessible complexity: enabling rich scenes on limited hardware
- Energy efficiency: critical for standalone headsets
- Scalability: the same content can run across different hardware tiers