High-Detail Cityscapes Generated from Satellite Data

During the production of one of my VFX animations, which was based on a real-world location, I developed an end-to-end pipeline that transforms raw geospatial data into highly detailed landscape renders in Blender.

The workflow was tested using official open geospatial datasets from Spain, France, the United Kingdom, and Switzerland, working with raster resolutions ranging from 0.5 to 2 metres per pixel. It combines orthophotos with Digital Surface Models (DSM) to accurately recreate real terrain and surface features.

The pipeline automates the entire process, from downloading and preprocessing the source data to generating an optimized mosaic of image tiles that can be efficiently loaded into Blender. Splitting the terrain into chunks allows very large landscapes to be rendered without exceeding Blender's memory limitations, enabling the creation of scenes that would otherwise be impractical to handle as a single dataset.

Using only publicly available geospatial data, this workflow produces medium-distance landscape visualizations with a level of detail comparable to Google Maps 3D imagery, while remaining fully reproducible and independent of proprietary assets.

The complete workflow is available in this repository: https://github.com/EnriqueFuster/geo2blender.