Caching models on the client vs direct replication

Help and Feedback Scripting Support
#1

Currently working on a turn-based fighting game. When spawning enemies in and using meshes in animations, I have two ideas:

  1. Cache enemy models that will be used in the battle and meshes in replicated storage, so the client can easily clone them. Use a remote event to tell the client when and what to clone.

  2. Just store everything in Server Storage and parent a clone into the workspace whenever needed/send a remote with the objects in it.

What is more efficient?

On one hand, I’m worried about client memory usage, which I can see is an issue with #1. However, hopefully the amount of cached objects is optimized enough because only the necessary objects are cached. There should not be any bloat at all.

#2 I feel is a lot more data heavy, but saves on a lot of memory. I’m only worried that a laggy player or heavier background calculations may have perceived input lag. (player → server does stuff → player)

Thanks!

#2

#1 Sounds more efficeint as clients are handling it independently and it keeps input lag low which makes the players experience better

#3

I think you should use the second one. If you store everything on the server and parent a clone into the workspace whenever needed, there will be less lag.
Also, when you store everything on the server, the player will have a better gaming experience. If the client is lagging, the player does not have to worry about it (if it is the server lagging, then the player has to worry about it).