Reconstructing Submerged Ice Age Landscapes

This project grew out of a workflow I developed a few months ago for a separate animation, where I needed to reconstruct terrain elevation in 3D. From there, I wanted to push the same pipeline further by working with bathymetric data and exploring a bigger question: what might parts of the world have looked like from above between 10,000 and 2,000 years ago, a period often linked to stories of great floods and large-scale inundations?

Across many distant cultures, with no obvious connection to one another, we find stories of floods covering vast areas of land. Could some of those stories preserve distant memories of the end of the Ice Age and the sea-level rise that followed? Whether or not that is the case, the idea was compelling enough to explore visually — especially because there are still very few detailed recreations of these lost landscapes.

For these visualizations, I used high-resolution bathymetric and elevation datasets together with satellite imagery, at around 30 metres per pixel. The main sources included GEBCO 2025 and EMODnet for bathymetry, and Copernicus DEM GLO-30 and GISCO EU DEM for elevation. I first built a pipeline to download and prepare the data, but the original raster files were far too large to load directly into Blender.

The solution was to split them into smaller tiles and create a custom import workflow. Inside Blender, each tile receives its elevation data, satellite texture, and land–ocean mask automatically. Those masks also control displacement strength, allowing the seabed and continental terrain to be treated differently. The hardest part was not the rendering itself, but turning a huge amount of geographic data into something Blender could actually handle.