A Data-Driven Path Synthesis Framework for Racing Games

We propose a data-driven race path synthesis framework with controllability for synthesizing realistic race paths. We represent a set of collected race paths in a path database using a formation graph in the preprocessing stage. By traversing the graph according to a given constraint, we generate a list of race path clips and combine and refine them sequentially to synthesize new race paths. Since the synthesized path is generated from existing paths with quality guaranteed, the synthesized path also provides the same level of quality. As the experimental results and user study results show, our framework produces good results effectively and is suitable for real time applications such as horse racing games.