Abstract: In this project describe a hierarchical routine for technically originating game maps using two-dimensional Markov chains. Our techniques takes collection of game maps breaks them into small chunks and platforms clustering to find a set of chunks that correspond to high-stage constructions in the practicing maps. Procedures that produce pc game at ease need game enterprise knowledge. It takes some method to mechanically study game proposal information for stage proposal from gameplay videos. In future validate the learned proposal information can be used to originate sections of game levels. In this project explored statistical procedures that could lead to generalized technical map generators. It shows rich game proposal information can be automatically analysed from gameplay videos and signified in reproductive problematic replicas. It takes Contra game for evidence. In the gameplay videos, it evaluates the method on a measure of execute ability and Quality to the real stages signified.
Keywords: Multidimensional Markov Chains (MdMC), Markov Random Fields(MRF), Hierarchical MdMCs, Procedural Content Generation.