31d1c0c112
- Move all demo projects that don't require Forward+/Mobile-only features to the Compatibility rendering method. This improves performance significantly on low-end devices and ensures visuals are identical to a web export of the demo. - Set deadzone on all inputs to 0.2 for better gamepad usability. - Remove reliance on `default_env.tres` to use built-in Environment resources in the main scene instead (which follows the preview environment workflow). - Remove notices pointing to GDNative or VisualScript, since both were removed in 4.0. - Various bug fixes and usability tweaks to 10+ demos.
Occlusion Culling and Mesh LOD
This demo showcases the use of occlusion culling and mesh level of detail in a 3D scene.
The demo contains 1,024 identical rooms in a 64×64 grid. The room geometry is used as a basis for a baked OccluderInstance3D, which allows each room to be culled if it is hidden by another room's walls.
Occlusion culling can provide a performance benefit in draw call-bound scenarios. However, since the Forward+ backend uses a depth prepass, the shading cost is already reduced by the depth prepass. As a result, occlusion culling will not always result in a tangible performance increase, especially in outdoor scenes that have fewer occlusion culling opportunities. Indoor scenes with no DirectionalLight3D casting shadows are a best-case scenario for occlusion culling.
The blue spheres make use of automatically generated LODs, which reduce the number of triangles that need to be rendered each frame by the GPU. This provides a significant performance increase in scenes with complex geometry. In most scenes, this provides a greater performance benefit compared to occlusion culling.
While running the demo, you can toggle the use of mesh LOD and occlusion culling to view the performance difference these features make in this project. Results will vary depending on your CPU and GPU model.
Warning
If you are using a engine build that is not fully optimized, you may notice that enabling occlusion culling decreases performance. This is because occlusion culling is a demanding process on the CPU, which needs all the build-time optimization it can get.
Official builds are fully optimized, but self-compiled builds are not fully optimized by default (use the
optimize=speed use_lto=yesSCons options).
Language: GDScript
Renderer: Forward+
Screenshot
