CityDreamer_云技派,专注ai人工智能! 140

CityDreamer, a 3D city streetscape generation model specifically designed for cities, generated by AI.

In recent years, extensive research has focused on 3D natural scene generation, but the field of 3D city generation has not received much exploration. This is due to the greater challenges posed by 3D city generation, primarily because humans are more sensitive to structural distortions in urban environments. In addition, generating 3D cities is more complex than generating 3D natural scenes, because buildings, as objects of the same category, exhibit a much wider range of appearances compared to the relatively consistent appearance of objects such as trees in natural scenes.

To address these challenges, we propose CityDreamer, a composite generative model specifically designed for unbounded 3D cities. CityDreamer separates the generation of building instances from other background objects such as roads, green spaces, and water areas into different modules. In addition, we construct two datasets, OSM and GoogleEarth, which contain a large number of real-world city images to enhance the realism of the layout and appearance of generated 3D cities. Through extensive experiments, CityDreamer has proven its superiority over state-of-the-art methods in generating various realistic 3D cities.