Download

What is the right way to model walls for an indoors level?

I am trying to port an architectural visualisation scene designed for static rendering from 3ds max into UE4 and I am not sure what is the correct way to handle the walls. The structure of the walls in the building is a bit odd (see image below). In the 3ds max scene the walls are simply modelled as a single mesh. That does not seem to be ideal for a UE level because the level needs to be navigable using a first-person controller.

I can think of 3 ways to handle this:

  • Use the mesh as is with complex collisions
  • Decompose it into convex fragments in 3ds max
  • Recreate it from scratch in UE editor using brushes

I was wondering if there is a recommended way to approach this?

In particular I am wondering if I try to decompose the mesh into simpler meshes, how do I make sure that there textures and lightmaps are consistent between the fragments?

3674f4859efbc6a7b80595bfa51944e5211239a9.png

You can extrude the spline out to create your walls, after that you would need to break it down into a number of separate meshes–each static mesh in UE4 will get its own lightmap, so for something like walls you’d need to separate it so more lightmaps can be used. As far as collision–best would be to create collision meshes in your 3D program and export to UE4.

As far as texture maps and separate meshes, you can setup your material UV’s before you separate the walls, and then after you break it apart you would create lightmap UV’s for each one. Luckily, with walls you can put lightmap seams in corners and it’s not a big deal.

But… that’s a screenshot from blender…???

Generally you would create walls in a game engine as boxes which you then link together. You could do this (with most of the walls being based on the same simple box and the oddball areas being separate entities) but it really only works if all the walls have the same basic material. But, if your walls have different materials (like something that is spread across a large area or several of your boxes) then you would have to break your texture down by wall and make sure the right one was on the right box. You can make all the walls one object but then you’ve got a texturing headache again, now you have to UV map a huge complex model. I’m not an Arch Viz guy but the first method makes more sense to me as it requires only a few boxes. Also the larger areas (where the walls seem really thick) don’t actually need to be that thick. Using the multiple box method would allow you to build what looks like what you have but is actually made up of the smaller boxes (assuming there’s no need for the client to actually enter that space).