I have to admit - I’m old school. Back in the day of the Quake engine and its variants, it was standard practice to break a level up using hallways that zig-zagged a lot to limit visibility in order to keep the number of polys on the screen as low as possible. This was a result of the engines being heavily dependent on CSG/BSP/VIS techniques common at the time. Flash forward to U4. I’m not familiar with how U4 does its VIS and spacial partitioning. So my question is what are the guidelines for creating multiplayer maps in U4? Just how open can you make them? Are there suggestions for making VIS work efficiently or are hallways still the way to go?
I could be wrong but since no one else is offering opinions. I would say it usually done with LODs or fog.
I’m pretty sure you can do it the same way, meaning you can have a character walk into a volume and that volume loads whatever comes afterwards and unloads the part that is behind the player. There’s a sample of this in the Contents Example Project. How you would handle this for multiplayer I don’t know.
What kennyrosenyc is talking about is called Level Streaming and is a definite option for single player but I feel like it isn’t a viable option for multiplayer but I could be wrong. It isn’t a topic I have explored too indepth.
In regards to your questions: Are you looking for an answer on how the rendering pipeline works or are you looking for more of a design aesthetic answer about limiting polycounts? I feel like there are a lot of ways to answer your questions and I am not sure which one in particular you are looking for.
Are you sure about making a level mostly out of static meshes? That makes life a whole lot easier. But it leaves me wondering once again how U4 manages to cull visible geometry without use of a BSP tree. I’ll assume an octree or maybe a kd-tree for now. It seems odd that brushes are out of favor when much of the editor is built to use them. I just hate the idea of making the world’s greatest super level only to find it is too slow to play on.
Test it for yourself. Make a complex level, full of long sight lines and complicated geometry. Test frame rates. (outside the editor) Then build chunks of it into static meshes and repeat.
Performance by gut feelings and hearsay will bite you hard. Take advice as a starting point to determine where to first focus your efforts, and then test and measure to make sure there’s a difference.
If static meshes are the new way to go, then hurray! Blender is lot better at creating geometry.
I was thinking about creating the major portions of a level using static meshes, but a question pops up: Does this mean you need to UV map everything?
It depends. When you just use simple cubes with 1 colour and without texture you dont have to uv map it. Otherwise you will have to create uv maps
OK. So let me ask this: if you make a room with 4 walls, does that mean you would typically need to make 4 objects? In other words, how much can you cram into a UV map before you run into trouble with the lightmap?
You can build it with 1 mesh -> modular
You create separate uv maps for your meshes and they dont affect the lightmap, because you have to create a own one for them: http://www.worldofleveldesign.com/categories/udk/udk-lightmaps-01-basics-and-important-principles-for-creating-using-lightmaps.php