Hi everyone,
I’ve been spending some time looking at performance optimization in larger Unreal Engine projects, especially environments that make heavy use of repeated meshes, detailed assets, and streamed areas.
One thing I’ve found interesting is that performance problems often don’t come from a single expensive asset, but from many small rendering and loading decisions adding up.
Some areas I think are particularly worth paying attention to are:
- Using ISM/HISM where many identical meshes are repeated
- Checking draw-call counts before focusing only on polygon count
- Using World Partition and Level Streaming carefully rather than assuming streaming automatically improves performance
- Reviewing LOD/Nanite settings for assets that appear frequently
- Profiling CPU, GPU, and memory separately before trying to optimize
- Testing packaged builds, since editor performance does not always represent the final game
I’m curious how other developers approach this.
When optimizing a large UE5 environment, what usually becomes your first major bottleneck: draw calls, GPU cost, memory, streaming, or something else?
Also, are there any profiling tools or workflows inside Unreal that you consider essential before making optimization decisions?
Would be interested to hear approaches from people working on larger production projects.