If I want to perform visual perception detection for multiple AI agents using object occlusion, how would this scale? Up to how many enemies would this be performant to?
Hi Mortaeus,
Dynamic occlusion (Hardware Occlusion Queries) are the most expensive method of occlusion in the engine since the query has to be sent to the GPU and read back again one frame later.
Use the command `Stat Initviews’ to monitor visibility/occlusion culling performance and counters to see what is happening in your scene with intent to keep Occluded Primitives, Visible Dynamic Primitives, and Occlusion Queries as low as possible while maintaining the performance you’d like.
Use the Desired Max Draw Distance in the level details to control the max distance these actors should be drawn. This should be able to be set in any Blueprints as well. This can help not render actors that are at far enough distances they shouldn’t be rendered.
I hope this helps.
Tim