unity-performance

Use Unity's Profiler to measure frame time, draw calls, memory usage, and GPU performance. Profile your scene during gameplay to establish a baseline, then identify which subsystem (rendering, physics, scripts, or memory) is causing frame-rate issues. This reveals where optimization efforts should focus first.

Draw-call reduction combines batching (static and dynamic), GPU instancing, and SRP Batcher. Static batching groups unchanging geometry; dynamic batching combines moving objects with similar materials; GPU instancing renders identical meshes efficiently. Choose the method based on your scene's object count and movement patterns.

LOD (Level of Detail) reduces mesh complexity for distant objects to save GPU cost. Occlusion culling skips rendering objects hidden behind other geometry, reducing draw calls and fill-rate pressure. Combine both techniques: LOD for distant or peripheral objects, occlusion culling for densely occluded environments like indoor scenes.

Yes. Object pooling reuses instances to prevent allocation spikes and GC pauses. Manage asset memory by unloading unused textures and meshes, using compressed formats, and streaming assets on demand. Profile garbage collection separately to identify allocation hotspots in scripts and reduce frame-time variance.

Core profiling techniques apply universally, but optimization priorities differ by platform. Mobile prioritizes draw calls and memory due to GPU and RAM limits. Desktop allows higher complexity but demands consistent frame rates. VR requires 90+ fps and low latency, making frame-time consistency and GPU-bound optimization critical.

Static batching combines fixed geometry once at build time, zero runtime cost but uses more memory. Dynamic batching combines moving objects at runtime, lower memory but higher CPU cost and limited to small meshes. Use static batching for level geometry; dynamic batching for small, frequently moving objects.