Example-based model synthesis

Model synthesis is a new approach to 3D modeling which automatically generates large models that resemble a small example model provided by the user. Model synthesis extends the 2D texture synthesis problem into higher dimensions and can be used to model many different objects and environments. The user only needs to provide an appropriate example model and does not need to provide any other instructions about how to generate the model. Model synthesis can be used to create symmetric models, models that change over time, and models that fit soft constraints. There are two important differences between our method and existing texture synthesis algorithms. The first is the use of a global search to find potential conflicts before adding new material to the model. The second difference is that we divide the problem of generating a large model into smaller subproblems which are easier to solve.