Large v Small Meshes Performance

[anonymous_user_7995904c]

I have large buildings that are the focus of my scene. I would really appreciate some feedback and advice on how performance would be affected by having each building as its own large mesh versus each building containing many small meshes.

From my understanding, large meshes would not be efficient for background objects because even if only one part of the building is seen the whole large mesh will have to be rendered. Since the large meshes I am considering would be the focus of the scene most of the mesh will be seen at all times. Also from my understanding, draw calls will decrease if there is only one large mesh. Would appreciate any advise on things I should consider to best optimize performance. Thanks!

[Rene_Zwaan]

Hey Apotdar,

Draw calls are often a lot bigger problem then polycount. This does not mean that poly count doesn’t matter.
Especially for a background object.

I would take a look at the building at it as a whole and remove parts I know the player will never see.
Do they never see the back? Remove the back.
Do they never see the table inside? Remove the table inside.

After removing all the parts I know the player will never see I will place the object in my level and reduce the polycount.

I have been able to sometimes reduce it to under 10% without being able to see a change in-game.

Also, check the number of materials you have on your mesh. Each material on a mesh adds a draw call.