AlphaModelRoadNetwork-cpp

An implementation of "A model for the generation of road networks" in C++

AKA the alpha model for procedural road network generation

Original paper

Building example

• Navigate to example
• Run scons
• You can run scons debug=1 if you want debugging symbols and no compiler optimizations.

Visualizing example output

• Navigate to example/bin
• Run example
• Run python ../visualize.py
• Enjoy the road.

Usage

Create an AlphaModelRoadGenerator object, with your given cities.

    // Grab  some cities
    std::vector<RoadGraph::Vertex> cities;
    cities.push_back({.point = RoadGraph::Point {.x= 0.5, .y = 0.5}, .mass = 0.26});
    cities.push_back({.point = RoadGraph::Point {.x= 0.86, .y = 0.85}, .mass = 0.50});
    cities.push_back({.point = RoadGraph::Point {.x= 0.16, .y = 0.9}, .mass = 0.45});
    cities.push_back({.point = RoadGraph::Point {.x= 0.1, .y = 0.25}, .mass = 0.7});
    cities.push_back({.point = RoadGraph::Point {.x= 0.6, .y = 0.7}, .mass = 0.4});
    cities.push_back({.point = RoadGraph::Point {.x= 0.23, .y = 0.2}, .mass = 0.1});
    cities.push_back({.point = RoadGraph::Point {.x= 0.15, .y = 0.13}, .mass = 0.5});
    cities.push_back({.point = RoadGraph::Point {.x= 0.88, .y = 0.3}, .mass = 0.25});
    cities.push_back({.point = RoadGraph::Point {.x= 0.1, .y = 0.5}, .mass = 0.35});
    cities.push_back({.point = RoadGraph::Point {.x= 0.9, .y = 0.43}, .mass = 5.0});
    
    AlphaModelRoadGenerator road_generator = AlphaModelRoadGenerator({}, cities);
    
    // Run your generation
    AlphaModelRoadGenerator::RoadGenerationOutput output;
    road_generator.generate_roads({}, output);

    // Your roads will be in output.edges, with output.vertices being your vertex positions

Settings

Graph settings

• bounds_*: The bounds of the area to generate a road in
• dummy_point_count: How many dummy points to generate to use as discrete points to build roads through

Generation settings

• alpha Alpha value, paper recommends 0.7 for human-like generation, the lower the more tree-like it is.
• road_straightening_factor: How much to straighten the roads after generation, gives a much more driveable result.

Potential improvements

• The ability to return complete road paths with road weights
• The ability to sample a heightmap
• The ability to reject edges in more complex ways (such as rejecting edges that cross over city boundaries)