Hello All,
Looking at the newly released Stalker 2 ZoneKit I noticed that they appear to utilise custom prefab actors and tools, and all of their actors in the actual game world are packed into API’s (ActorPackedInstances)
It made me wonder why? They use a very very heavy modular workflow and all these API’s are essentially grid cell based packs of instanced static mesh actors, eg one for architecture, one for foliage etc etc.
It makes sense to pack all the architecture prefabs per cell into singular large ISM objects which to my knowledge cant be achieved with level instances, but these prefab actors don’t seem to be as easily updatable as a level instance, nor can I tell whether their workflow is destructive or not.. I.e. if they want to change the design for House_A which exists 100 times across the map, I’ve no idea how they regenerate those API’s, whether thats automatic or not, all I can say is the ‘source’ assets don’t appear to exist in the world map that the API’s build from.
Hopefully that makes sense, but in short:
TLDR: What exactly is best practice for large open world performance? Level instances seem lacking, Stalker 2 using a custom solution says a lot in my opinion?
J