Crypto Cleveland

Generating Terrain

I’ve been screwing around trying to generate some terrain in Blender using Bump, Normal, and Displacement Maps.  

I tend to lump these modifiers together, the common thread being that you use a 2d image to create 3d geometry, or the illusion of it; but they are technically all different.  Bump maps and Normal Maps are created by analyzing a 3d model in order to create a 2d texture.   A Displacement map is the reverse where you use a 2 dimensional image to displace vertices on a polygonal mesh.

Here is a brief tutorial that explains the concept of Normals - http://www.youtube.com/watch?v=U7PQGgz1RII

Now topographic data is actually pretty hard to come by, at least the pickings are slim when you talk about what is freely available for use on the web.   I think the reference image I came across is originally from Rails and Trails.com

This map is from about 1918 or so, but the infrastructure is not substantially different from what existed in 1896.  As you can see the imagery is not the greatest of quality, especially when you consider that 9 square miles of the city are represented in this thumbnail and my project focuses on 1 square mile of downtown at the center right of the image.

Basically I had to generate my own terrain map by tracing the little red lines you see in the map above, filling the boundaries with progressively lighter shades of grey, and then smudging the borders so that you get the image below.  

Once that is achieved (It takes a good bit of photoshop/gimp knowhow)  you can use the map to displace vertices on your mesh… how detailed the terrain is really depends on how many vertices you have.

This is a “Bump Map” applied to a 2 dimensional plane to give the illusion of depth

This is 3 dimensional geometry modified with a 2D “Displacement Map” plus a subsurf modifier to smooth the angles (it would otherwise be indistinguishable from the bump map).    What you see is a plane with 7 subdivisions or 16k vertices.  

This is that same geometry with a texture applied so that I can correctly align streets etc.  At this stage the goal is to keep the overall number of vertices/polygons low. If you need more detail you can always do some “digital earthmoving” later on.  Remember that most of the terrain will only be visible in the foreground or background. 

For example: If I wanted to do a birdseye view illustrating the activity along the river and waterfront, I might frame it something like this…. The focus of this shot would be the bridges in the flats, otherwise the bulk of the image would be the main model of the downtown core, and Lake Erie, which could be seamlessly composited with a contemporary photo.