Urban growth has been widely studied and many models (in particular Cellular Automata and Agent-Based Models) have been developed. Most of these models rely on two representations of the geographic space: raster and vector. Both representations have their own strengths and drawbacks. The raster models are simpler to implement and require less data, which explains their success and why most of urban growth models are based on this representation. However, they are not adapted to microscopic dynamics such as, for example, the construction of buildings. To reach such goal, a vector-based representation of space is mandatory. However, very few vector models exist, and none of them is easily adaptable to different case studies. In this paper, we propose to use a simple raster model and to adapt it to a vector representation of the geographic space and processes allowing studying urban growth at fine scale. Both models have been validated by a case study concerning the city of Can Tho, Vietnam.