Server-Side Anti-Cheat: Fog of War and Occlusion Culling Vector Verification: Advanced Implementation and Real-World Architecture

[Austin Cox]

I’ve been architecting a server-authoritative visibility system for our competitive title and hitting some real friction with the occlusion culling pipeline. When the server determines a client’s frustum and line-of-sight, how are you handling the culling of occluded entities—pure server-side spatial partitioning with a broadphase octree, or are you pushing visibility decisions down to a client-side query that gets validated server-side? The concern is that if the client is doing its own occlusion test and just suppressing the draw call, a determined cheat can bypass that by reading the scene graph directly. What’s your take on keeping the visibility determination purely server-authoritative while still minimizing the bandwidth hit from sending entity state for off-screen actors?

On the performance side, what kind of CPU overhead are you seeing when running these visibility queries on every tick for a 128-player server? Are you batching the occlusion checks or interpolating between frames, and how does that impact the render latency budget on the client? Also interested in how you’re handling fast-moving targets where interpolation creates window where an entity briefly becomes visible server-side but the client hasn’t updated its occlusion state yet—anyone found a clean solution for that race, or is delta-time clamping the standard approach?

"0 dB Idle Fan Profile: Fans completely stop below 45°C GPU temp"

[Author]

Posts