3D PROCEDURAL MESH GENERATOR

Abstract

Activation of 3D rebuilding from 2D is an interesting future service for small groups of artists or independent game developers who need a lot of 3d assets for their project in less time. In order to give a realistic notion of what the technique used in for proposed project is capable of, the Convolution Neural Network (CNN) is examined in different aspects. The task is to create a CNN that can be used in 3D to examine and scale the topology of a model from a 2D environment and mapping it into 3D model. The main areas for which CNN is optimized are quality, speed and simplicity. While the output resolution of the quality 3D object matters in this field, the speed also matters which is measured by the number of seconds to complete a rebuilding and  another aspect i.e. simplicity is obtained through Machine Learning (ML). Enabling this can facilitate the creation and development of 3D games.