Would Combining hundreds of Meshparts make a Large model more performant?

Help and Feedback Building Support
, ,
1

I am building a massive tower out of mesh-parts (most around 1k tris). It is huge, 550 studs tall and 150 studs in diameter. It consists of just under 3000 individual mesh-parts.

I wonder if anyone knows whether or not combining the meshes into much fewer meshes would make the tower more performant. Combining the meshes would mean they are made of the same amount of triangles but not as many individual mesh-part objects. It would also be pretty time-consuming which is why I thought I would ask around first.

Does anyone know if this would actually improve the performance of the tower? I assume it would considering there would be much fewer objects and therefore much less data being manipulated when it is loaded, rendered, moved, etc.

I was able to combine 250 mesh parts into 30 in one instance, I would imagine this would help the performance of the tower significantly if I did it to all of the major groups.

1 Like
2

it might help with creating the collisions (probably as it only has to render one mesh and not 300 individual collision boxes) but you should use the boolean modifier if you are using blender so that there are no unnecessary faces that have to be rendered and of course don’t turn on double sided

1 Like
3

Mind if I sneak a picture of it? If I see what it looks like now, I can base my thoughts off the geometry itself

4

Screenshot_218
Screenshot_2181340×969 114 KB

Screenshot_219
Screenshot_219416×689 466 KB

1 Like
5

In your case, the color scheme is so similar you could megatexture the entire tower one texture, and try to combine as many modular parts of the tower into 10k meshparts as possible. Megatexture, and optimizing for lower polygon / less instance meshparts