I had exactly the same thing. Basically there is a cost to getting the instancing ‘off the ground’. Until you hit that threshold, placing static meshes will perform better than ISMs. If you try and populate a large area with static meshes, I think you’ll see the difference…
Hello everyone.
I recently noticed a strange behaviour that I don’t fully understand. Lets say I have 1000 trees in my scene painted with the foliage painter. As far as I know the foliage tool uses hierarchical instanced meshes for a better performance. But when I place the trees manually as individual static mesh actors the performance is much better compared to the foliage tool. That doesn’t make any sense to me.
Then I made a test bluperint that spawns a grid of trees (lets say 30x30=900 trees) with a boolean to either add a static mesh component or a hierarchical static mesh instance. And again the same result. The difference is not as big as compared to the foliage tool but still I have a higher fps with just regular static mesh components. How?
Did anyone else noticed this behaviour? Any explanation why is an instance slower than an individual actor? Thanks.
does this apply to HISM? Are HISM good to use to make modular maps? I noticed I cant cull because if I cull 1 area it culls parts and pieces of my modular map.
so it makes no sense to use them is what you are saying? where is the threshold?
Frankly, I’m no expert on this, but had already asked the question elsewhere on AH, hence my answer. But, I would imagine that HISM have the same issue, the threshhold is probably somewhere in the ( very ) low thousands ( just off the top of my head ).
If you have a map large enough to need streaming ( is that what you mean by modular? ) then I would use instances 