So I am in work to import to Unreal a huge model of the Colosseum. It is modeled in Blender and has about 1 million tris.
It is being created with about 80 small assets (arches, stairs, etc). Final model uses copies of that assets with a total number of about 900 objects. Only one quarter of the building is being modeled, the three other are created using mirror modifiers.
I would appreciate any idea for doing the import as fast as possible, and for any suggestion in performance improvements.
These are some ideas I have for importing:
Exporting each asset to FBX, importing in Unreal. Drag and drop each asset using objects in Blender as info for location, rotation and scale of objects. This task requires a lot of time because even reducing the building to one quarter, still it means to drag and drop about 200 objects in Unreal.
Exporting selections of objects in FBX, then importing to Unreal. Problem here is that Unreal deal with the content of each FBX as a single mesh, so how to export there UV lightmaps without overlapping issues is something I do not see clear.
Creating an addon in Blender to export the info of position, rotation and scale of each object, and then create a plugin in Unreal to read that info and creates the model on-the-fly.
Regarding performance the main idea is to create different LODs for each object, but not sure if there are better ways to do this. For example, not sure how to improve object occlusion in a buildings that has a lot of sightlines.
I guess the best way woul be, to build up the Colosseum in UE4. Import your 80 small meshes and go for it
If there is a better solution on Import a Object like this, i would know it too.
The best way is to create modular assets which you place into the UE4 + you build the building there. -> you can use meshes over and over again; when you use single parts you can cull them + you can add lod’s
, could you expand what you mean with “cull”. I suppose you mean occlusion. Is there a method for improve the object occlusion in UE4, a kind of objects that can be added and tagged as occluders?
I am considering some ideas, but not sure if UE4 supports them. For example, one thing that could improve a lot the performance is using LODs that group objects together, making that several objects in LOD0 become only one object in LOD1. I think 900 objects is very high number and they are going to suppose extra time when UE calculates the object occlusion or even the number of drawcalls. Any idea of how UE could support LOD groups?
If I were to need to process a model like this, I would export all your little assets you’ve made as suggests. Then write a script to emit a t3d file from blender that describes their placement. That would allow you to reconstruct the thing if you make tweaks to the placement in blender.
Place a few static meshes in a scene in unreal, select and copy them to the clipboard.
Paste the contents into your favorite text editor and you can see that it’s a pretty simple format.
I often import larger assets then this, and just select all the objects and drag them into the viewport as one, they line up perfectly.
They are exported from though, but should be the same from Blender I guess.
I got my own Colosseum mesh which I made a few years ago into Unreal Engine 3, and made a map out of it for Chivalry: Medieval Warfare (now the map will be released as an official map) by breaking the Colosseum up into modular components since game engines will have an import limit on polys.
I can’t remember how many polys are on my mesh (lots), but I just broke everything up into chunks (I think it ended up being about ~30 modular parts), I centered the pivot point of each module to 0,0,0 in Maya, and for every piece I imported to the engine, I would just snap to the same location as every other piece, so it wasn’t finicky to get all the pieces in the right locations. The hardest part (which was just tedious) was breaking everything up into those chunks and making sure that they were under the import limit.
I didn’t use culling (since UE3 was more than capable), but I believe you can create your own occluding mesh if you wanted to, so that while you are inside of the arena itself, you won’t be able to see all of the interior (underneath the arena and stands).
You can see the result of the work here. (Bare in mind it was a failed attempt at making the Colosseum (I missed one measurement and had to improvise)) and in no way is final. I also had to scale the Colosseum up a little because players couldn’t fit inside the smaller corridoors due to the crazy big player collision meshes in Chivalry.
