How to export?

I’ve been working on an online multiplayer game for some time now, and everything’s going well. At the moment I’m implementing server-side collision detection using the bullet physics engine. The map was created in blender, but I need to export it to Collada so that my server can import it as an object.

My main problem is that in order to have collision detection on other objects(trees, rocks, etc.), they need to be exported with the map. My main concern is all the extra faces being exported(leaves of the trees, faces added for detail, etc.).

With all these faces being included, I’m pretty sure that my server will crash. Basically, I’m wondering if anyone here has any ideas on how to reduce the faces being exported. I know it’s not a trivial problem, but any brainstorming would be appreciated.

So… are you using the BGE or not? It doesn’t sound like it. If you aren’t, then this probably is the wrong forum.

Can’t you make those “extra faces” seperate from the main mesh? Or just create a collision mesh for each object, and only export those?

Why server-side collision detection? Your client lags, and he’s going to run right through a wall…


We are using the BGE for the client, and I’m not sure this topic would fit anywhere else in the forums. The problem with making the extra faces separate from the main mesh is the vastness of the project. While that would be possible, I’d rather think of a more automated way of doing it.

One way of doing it would be to use your idea of separating the faces and using those two meshes(main + extra) together as the mesh that the client will see. I would then write a script that would go through all the objects and determine which ones to export based on their names.

Why server-side collision detection? Your client lags, and he’s going to run right through a wall…

I’m doing server-side collision detection to prevent people from walking through walls. With server-side collision detection I can make the server fully authoritative of the player’s position.
While blender has collision detection of it’s own, there would be nothing to stop the client from saying “I move 5 steps this way and end up here”, even though in reality he’d have to walk twenty steps around a wall to reach that spot.

It’s all about security.