Case Study - 3D City Modeling
Updated: Dec 18, 2020
This project entails a semantic 3D city model of a city with around 30,000 inhabitants based on datasets provided by our client. The provided input data contained the point cloud of the exterior of about 3000 buildings. The digital building model was supplemented with detailed 3D laser scan data. This project included detailed and accurate 3D modeling with texture and layers of residential buildings, commercial buildings, roads and a few heritage and monumental buildings of the city.
The building models are created from point cloud data in conjunction with height information, from 3D shape files, and various 3D model formats (e.g., .3ds, .x3d). Besides this, multiple layers of vector and raster data are added as terrain textures and labels or (3D)-symbols are added based on point information.
One of the major challenges we faced while executing this project was its visualization, semantic rendering and maintaining the quality. Visualization acts as a challenge because the components of 3D city model are mostly from different data sources.
Another challenge posed was the non-availability of proper texturing file and blue print. Since there are some restricted areas in the city, the laser scanning data and photos could not be captured properly.
In the first attempt of 3D models visualization, small objects like architectural models were displayed in wire-frame format or using CAD packages, while terrain models were visualized in perspective wire-frame models with draping of orthophotos or orthophoto maps.
We created a single texture image using reference images. Once the texturing was done, we carried out the layerization. In order to ensure quality, continuous QC and QA process was undertaken. We also took reference from Google map and Google images to cover the restricted areas.
This project provided an intuitive media for the visualization and comparison of urban design proposals. Users could interactively navigate through the 3D city model and attribute information attached to buildings and vector data could be queried. In this project, elevation values of buildings and names of blocks were attached as attribute data to the buildings.