A question regarding modular buildings

synergy11

Good day.

I am having trouble wrapping my head around when I should be creating fully modular buildings or "hero" buildings.

http://lazyphotog.files.wordpress.com/2010/04/saturdayapril-118-edit.jpg

Taking into account only the bottom half of the building, how would I create a modular layout for the main wall and columns. This building design has them looking all like one piece. And I want it one piece in UDK so I can use texture blending and vertex painting to really push the look of this building wall texture. And If I break the main wall and columns up into different pieces the vertex blending won't look right over the models edges where they meet. Or will they?

However everywhere I look online for modular buildings, it always has people breaking everything up into little chunks. WHich is great when the building design can support that workflow. But in this example I can't see how it would.

Obviously I can create the windows and trim parts as seperate models with one texture sheet.

Any help would be great on this matter!

Obscura

Well.Breaking the building into parts would break the flow of the vertex painting, and could easily cause lightmapping problems, so yes you are right. I would recommend to use modular pieces in your modeling software, welding when its done (welding the uvs too) and exporting it as one big mesh.You can use elements for things like windows,etc, but the walls would be better as one thing. Or a few. An another thing is a draw call question. As I know, if you use separated meshed,they will use separated draw calls. So welding and exporting it into bigger pieces would be cheaper.

synergy11

Thanks for the help.

I didn't know bigger chunks would use less draw calls, then instanced pieces.

Ark

Yeah having less pieces is better for reducing draw calls, but not so good for occlusion.

If you're building tall skyscrapers, then it's better to keep it sectioned in grouped stories then the engine doesn't need to draw the higher tiers of the building when the camera isn't looking up. Same applies to walls that aren't facing the camera, they can be unloaded once the camera can no longer see that particular side of the facade wall.

I would say it's a balance between making the building modular enough so pieces can be reused and keeping the model split up wisely for occlusion, but yet not going overboard on the number of pieces needed to complete the structure.

sprunghunt

this might help:
http://udn.epicgames.com/Three/ProceduralBuildingsTutorial.html

Module size is all about how far away from the player the object is. If you can walk right up to the bottom of the building then you'll probably want many small bits. If it's only a background building then you can probably just make it one big piece.

synergy11

Yes this is a building the player can walk up to.

I don't want to compromise the ability to blend textures with vertex painting. So I can't break it up into smaller chunks. (At least the main brick wall part anyways)

For the occlusion problem, I can separate the top part of the building and the bottom and middle sections easy enough. I will probably have to separate the front where it curves as well.

Will this be a good trade off for quality and performance?

Thanks again.