Hi,
I’m making an fly through in unity for an oculus presentation. The fly through will go through my virtual building which is unfortunately very heavy in terms of mesh faces.
Loading the whole building at startup will murder my framerate so I was thinking of splitting my building across several scenes then additively loading and unloading the pieces as the camera moves through.
Is there any performance issues that are likely to arise from this? Or is there a better way?
I guess depending on the size of the scene loading it’s going to stop the flythrough for a second or two on loading.
Thanks,
Chris
There’s definitely a few ways to segment a world and stream it in on demand, but streaming may not be the solution you need in your case. You say your scene is vertex heavy. Are you sure this is a memory issue? Memory usage is not usually the cause of low frame rate. It sounds like your just drawing too much. How many triangles do you have on screen? Have you looked into LODs, occlusion culling, deactivating objects which you know don’t need to be drawing, and things like that?
Hi dj,
I do not think it’s a memory issue as much as it is drawing way too much. I have a ridiculous amount of triangles (several million) due to my modelling workflow, significant under optimisation and lack of skills.
I haven’t heard of LODs or occlusion culling. Reading though, I think you may be right.
Sorry if I sound like a fool, this is my first time using unity. I have a background in architecture.
I have implemented some occulsion culling, a secondary question I have is why do I get 800k tris even when nothing is inside the camera view?
You should defenetivly use occlusion culling!
occlusion culling in short: What is obstructed and can not be seen will not be rendered.
LOD / Level of Detail:
If you are not sure what’s get rendered , use the frame debugger to figure it out.
Fantastic! These two things have helped so much.
Glad I asked before going down the rabbit hole of splitting everything up! Thanks
