Unreal Engine 4 - Repeating meshes VS lightmap quality.
polycounter lvl 12
I've got a mesh that has to repeat about 32 times, a large cylinder like structure, so every piece has the same Materials and UV data.The problem I'm running into is my Lightmap UV in Unreal:
Would it be better to just have one piece of the mesh repeat 32 times INSIDE Unreal with Good lightmaps. So a tiny portion that I just keep duplicating.
Or Combine let's say 8 pieces into a single mesh and have that mesh only duplicate 4 times but with really packed and kinda shitty light maps.
Is there any downside to having that many static meshes in one scene? other than piecing it all together.
Would it be better to just have one piece of the mesh repeat 32 times INSIDE Unreal with Good lightmaps. So a tiny portion that I just keep duplicating.
Or Combine let's say 8 pieces into a single mesh and have that mesh only duplicate 4 times but with really packed and kinda shitty light maps.
Is there any downside to having that many static meshes in one scene? other than piecing it all together.
Replies
This would be easier if you showed a picture of the mesh in question, since a lot of it depends on the complexity and type of mesh you have.
Some Articles that might help:
http://gamedev.stackexchange.com/questions/81688/road-network-geometry-in-unity-one-big-mesh-or-a-bunch-of-smaller-meshes
If anyone could help any further, Here is the mesh in question
As you can see its just a repeating mesh that I could divide up further into 8 parts or less, At the moment it's 1/4 of the entire part.
However this depends on how much memory your first object takes up. So there is a threshold where it's better to join objects together. And there's another threshold higher up where it's better to have smaller objects. If your mesh isn't too high poly then I'd join them together.
This also applies to performance. Any platform can only render so many polygons. So if you have one large mesh joined together it's going to try to render the whole object. This can cause greater strain on the renderer as you won't get any of the advantages of having polygons hidden by culling out objects.
However rendering individual transforms also takes time. So having a lot of tiny individual objects in scene will also slow down the renderer. So there's also a lower bound for this case too.
In other words try to keep your meshes fairly similar in the number of polygons that they contain. You can vary up and down a bit but the more even it is the better. So simple geometrical shapes can be a lot bigger than complex shapes.