A Method for Automatic Generation of 3D Background Models Using Fractal Models and Genetic Algorithms

In recent years, 3DCG-related industries have grown rapidly and the demand for 3DCG content has increased. However, 3DCG production requires special skills and a lot of time. In particular, the creation of background data requires a lot of human resources at the production site because the data is huge and complex. Therefore, we propose an automatic creation system that generates 3D background data for 3DCG production. The proposed system consists of three functions: 3D model generation, generation parameter optimization, and fitness evaluation. The user inputs a landscape photograph as idea information into the system, and the system optimizes the generation parameters so that the 3D object shape approximates the input image. To evaluate the accuracy of the system's performance, we observe the optimization process and the 3D object output when a previously generated fractal image or landscape photos are input. Experimental results show that the system accurately optimizes the generation parameters and generates 3D objects that are almost identical to the shape of the fractal image. In addition, the parameters were also optimized with a certain degree of accuracy for landscape photographs for which no specific solution had been determined.