I am in the early stages of making a PVE game. My plan is to have all entities on the server be represented by a single cube for physics handling, and then the visible model and collider will be rendered on the client. What are some potential issues that may arise from this, security aside? This is a decision i will need to make early on. Designing it this way will work wonders for performance on the server, but if I figure out later on that this is inviable, it may mean having to rework a lot of systems in the game.
It is still rendered on client?
What are you talking about?
The only thing it may do is to optimize replication a little but it will 100% mess up all StreamingEnabled if you do.
Streaming wont be an issue, I can remove the client model and client collider every time an entity stops existing on the client.
This also wont just optimize replication, it should also optimize server physics in general. Think if there are 150 enemies existing in the game at once, that will be 150 moving parts the server needs to handle. Now imagine, each enemy has a detailed collider of ~25 parts each. Now the server has to handle 3750 loose parts. Since most of these parts are welded to the roots of their models, this will only be 150 physical instances, but its still 3750 individual colliders. Correct me if I’m wrong, but that should be a huge performance boost, right?