Need some guidance in how best to approach 3DSMAX modeling of large ARCH Building.

I have a very large two story building that is in REVIT. I am an experienced MAX user/modeler and I want to remodel this building completely in MAX for bringing into UE4 for my first Environment composition. How is it best to divide the geometry up so that everything falls onto grid? I just do not see how I can break this irregular shaped building up into modular pieces like everything I have read suggests? I cannot seem to get a pic uploaded as the forum app seems to error off every time…so I posted a link to the elevations.


Any advice or help greatly appreciated!!


Modular workflow doesn’t always work, and often times when your purpose is rendering rather than a game it isn’t a good option.
For your case, I would do the outside walls as one mesh, if you can then you can split them into multiple meshes at the corners (since each mesh can only have one lightmap you can get higher detail lighting by splitting them up so that they use more lightmaps)
For the windows, you can make those modular since many of them are duplicates. Try to keep things together that use the same material, but if you can split on corners like the walls then that can help as well.
Look into FBX scene import:
Scene import will allow you to place your meshes in 3ds Max and export the whole thing to a single FBX file, it can then separate each object out into the Content Browser in UE4 but you’ll get a Blueprint file that contains information about where they are placed which you can drop into the level and have everything placed where you want while maintaining instancing.

Thank you for the great info!
If I was going to start developing assets that look like this type of structure, for gaming, how is that done in general? Can it still be done in MAX or should I be looking to another package for that kind of modelling? I would really like to build a kit of modular parts so that I can quickly put together parts of a building.

Max is fine for modeling anything. For modular workflow there’s a couple things to consider–one is that each object increases the number of draw calls which slows down performance so avoiding adding more objects (even if they’re copies) will make things faster. The other issue is specific to UE4 in that the lighting gets processed on each individual object, and that means that indirect lighting will look slightly different due to things like the smoothing settings in the lighting. Because of that, if you have something like a flat surface and you split it into multiple meshes then each one will end up looking slightly different rather than looking like a single surface. Due to that lighting issue, it helps to try and keep surfaces as a single object as much as possible which can require making unique wall meshes. It’s not as much of a memory issue though to do that, since usually those meshes are very simple so it has little impact on memory to have unique meshes.

See anything wrong with this process?


For what it’s worth, I plan on doing a video log of the entire process once I have things sorted for anyone that is interested.

You’ll need to delete the polygons that cap the bottom and tops of the walls (they aren’t necessary since you won’t see them). If you’re doing the interior as well then you’ll need to split the interior walls off and then the same idea applies to keep them together as much as possible and split on corners if the mesh is too big (where you’ve reached the point where a large lightmap can’t provide enough detail because of too much surface area).
Another thing for the interior you’ll need to do is have the floor and ceiling meshes match the walls exactly rather than doing something like taking a large plane and just placing it over everything. This is also due to the lighting–parts that extend over the wall or appear outside will receive lighting or shadow that’s not on the inside, and since the lighting gets save to a lightmap (basically just an image of the lighting that gets mapped to the mesh) the pixels can often bleed onto the visible areas inside causing weird shadows and lights in the corners.
Things like props are what you’d definitely still want to do a modular workflow for, they work fine with lighting you can just copy and paste those around.

I haven’t tried the “full scene” workflow that darthviper107 suggests. But let me add a few comments because I love modules :wink:

I can see many objects that do repeat in that building. Smaller ones like the windows and bigger ones like the vertical elements with the little roofs. Also the entire building is symmetrical so in theory you only need to build the center and one side. You can mirror the other side.

Here are a few tips

  • The flat building walls and the flat roof parts are fairly simple so they can be larger meshes.
  • Most work will sit in the detailed ‘add on’ meshes. So you don’t want to copy those in max because if you have to change them you will need to change them in each copy.
  • to know where things exactly are meant to be you could import the entire building as a ‘helper’ mesh. Put it in a helper sub level that is easy to turn on and off. The building isn’t too large so it should be fine to have one huge mesh as a reference in a hidden level
  • as darthviper mentioned it is a balancing act between draw calls and number of pieces. If you have too many pieces with multiple id’s the draw calls can severely bring down performance. But in this scene it shouldn’t be too much of a problem unless you start building parts out of simple boxes. Try to combine as much as possible WITHIN the modules.
  • In games we used to split the entire scene into modular proxy meshes, place them in the level until the scene was complete and then improve all the individual pieces one by one (without touching their location in 3ds max).
  • Put the modules on location 0,0,0 in 3ds max before you export them. You can always leave an instance where it is meant to be in max. Otherwise you will have a hard time to move them around.
  • See static meshes in unreal as being the same as an XRef or instance in 3ds max. You place them in the scene and once you reimport a better version all instances placed in the scene will be updated. This seems to be common sense but I just want to mention it because some people seem to have difficulties with this concept as it is different than working in 3ds max.