Jump to content

Single Mesh from Clutter


photo

Recommended Posts

Hello,

 

Whats the workflow from creating a single surface mesh from a clutter object.

 

I want to scatter stones with clutter object.

 

Then I want to create a single object with a single surface (I guess thats the cheapest way to render it or?)

 

I do it :

 

a) create clutter

b ) export a mesh

c) create object > static mesh

d) export fbx

e) import into 3dsmax

f) collapse object

g) export fbx

h) import fbx

 

Alot of steps involved....

 

Thx.

Werner

Link to comment

 

 

Then I want to create a single object with a single surface (I guess thats the cheapest way to render it or?)

 

 

This depends on the use case. In case of 20 identical rocks on very small place (always in frustum) it is little bit cheaper, because it is one object instead of 20, but number of drawcalls will be same, because AFAIK Unigine batches same meshes with same material.

On the other side if we are speaking about 10000 rocks spreaded through several square KMs, it is FAR better to use clutter, because of frustum culling (frustum culling is done per object, not per surface) and clutter effectivity (it does not keep in memory 10000 instances, only the visible ones).

 

We are using same workflow as you described for optimization purposes of some structures. Usually houses built from many small parts. But never for clutters. As object grows bigger, it is more effecient to divide it into more piecies to use benefit of frustum culling. For example this building:

 

https://developer.unigine.com/forum/topic/3230-el-somni-quas/#entry21475

 

has 25M polygons and is made of 12000 instances of cca 40 unique objects (cca 45k surfaces together). Perfomance is fine, because of LODs, batching, oclussion objects and frustum culling. If you merge it into one object, it will create FBX with size 1,2GB (!!!) and framerate literaly 0 (you loose benefit of instancing, frustum culling and oclussion). So for close distance we keep it as is and for bigger distance we model manually one object LOD.

Link to comment

Then I want to create a single object with a single surface (I guess thats the cheapest way to render it or?)

That's not entirely true. In many cases having a cluttered object is more performance friendly rather than having 1 big single object. 

 

You can try to use MeshCombiner plugin that will bake all the selected ObjectMeshStatic objects into a single object, but you may not receive expected performance gain.

How to submit a good bug report
---
FTP server for test scenes and user uploads:

Link to comment
×
×
  • Create New...