Hello everyone!
I’m trying to do a archviz scene and I want to do a row of 3 detailed windows. This stack of windows will be in other places too, at the end it’s going to be a very populated scene and I want to build it taking care of the performance/drawcalls.
For testing purposes I modeled in 3ds max a field of 40 x 40 cubes with 100 subdivisions each one. I selected them and exported/import to unreal, the loading of all the unique cubes in content browser was slow of course. drag and drop into the scene slow too. Then I applied a simple white material on all of them and watch…about 28 fps on my computer.
Then I try it with importing one unique fbx cube, and Alt dragging and creating copies inside the scene until i get the same 40 x 40 and same camera position…about 28 fps too!! Does this mean that the engine doesn’t care about the source, fragmentation, batching… of how we populate our scenes or that I need something like 200 x 200 unique objects to start to see some fps/ms difference with both methods?
Importing all fbx cubes is good for me because it’s a true copy of my max scene, Unreal doesn’t have the same snapping and precision controls, I can’t rely only on alt-dragging in the engine.
Yes, probably the answer is to combine all the meshes to be one object in max (because “combine meshes” in unreal generate problems with Uv’s and custom light maps) But I want to know how the engine handles the above mentioned for other situations because combining into big objects cause constant drawing when you see them only a little bit.
Oh, and a second quick question if I may ask
If I want to apply exactly the same material for all the windows, it’s the same performance friendly/drawcalls to just apply the material to all of them or use a material instance?
Thanks a lot! Very good engine and community! I hope to contribute soon (with improved english too).
