Quick Answer: The Switzerland 3D terrain model at blocks.manifold3d.ch is a free download of swisstopo map and elevation data for the Jungfrau-Aletsch UNESCO area, about 2,400 km². The 1.1 GB ZIP is formatted for Prism and needs no registration. At true 1:1 scale, work in smaller tiles, because a full block-per-metre build of that area would contain billions of blocks.
Switzerland is one of the few countries where you can get official, high-resolution elevation data for free, and a project has now turned a slice of it into a block-by-block terrain set. The source page lists the Eiger, Mönch, Jungfrau, the Aletsch glacier, and the Oeschinensee among the features included. The idea is simple: real mountains, rebuilt one block at a time, at the same scale as the ground they sit on. The execution is harder than it looks, and most of the difficulty comes from one question: how fine should the blocks be?
What the Download Actually Gives You
The download covers the whole UNESCO World Heritage area of Jungfrau-Aletsch, which Switzerland inscribed in 2001. The source describes it as roughly 2,400 km² and credits the map to swisstopo, the Federal Office of Topography. Swisstopo has published much of its geodata under open terms in recent years, which is why a site like this can offer it free without a sign-up.
Keep in mind what a terrain download is. It's a height field, a grid where each cell has an elevation value, plus map imagery or colour data. It is not a model of rock, ice, or forest structure. A voxel build turns that height field into columns of blocks, and everything above the ground (trees, buildings, the lakes' depth) has to be added or approximated separately.
Here's the first thing to check before you download: the ZIP is 1.1 GB, which is large but manageable. The real cost shows up after you import it.
Why 1:1 Scale Changes Everything
At 1:100 or 1:1000, a mountain fits on a table or a build plate, and the block count stays within reach of ordinary hardware. At 1:1, each block represents one real metre (or one real half-metre, depending on your setup). The area stops being a model and becomes a count.
Here's the arithmetic. 2,400 km² is 2.4 × 10⁹ m². At one block per metre, the surface alone needs about 2.4 billion columns, before you add a single block of height. Even if you only use the top layer, that's a surface-only figure that most editors and renderers won't handle comfortably.
Here's how the surface column count changes with grid spacing. These are columns for the surface layer only, not total blocks:
| Grid spacing (horizontal) | Surface columns for ~2,400 km² | What it's realistic for |
|---|---|---|
| 1 m | ~2.4 billion | Single valleys or summit clusters, not the full area |
| 2 m | ~600 million | Sub-regions, such as one glacier basin |
| 10 m | ~24 million | Full-area overviews on a large machine |
| 25 m | ~3.84 million | Whole-area builds that stay manageable |
The numbers make the trade-off visible. Each step down in grid spacing multiplies the work by the square of the ratio. Going from 25 m to 1 m multiplies the surface count by 625.
The Counter-Intuitive Part: Finer Is Often Worse
The popular advice is to always pick the highest-resolution source. That advice is incomplete for block builds, and here's why.
First, blocks are visually coarse by design. A block at 1 m is already a clear square when you look at it from a few hundred metres. Feeding it a 0.5 m height field doesn't add the detail you'd expect, because the blocks round the surface back to a grid. You pay for resolution you can't see.
Second, finer grids produce noisier surfaces. A high-resolution elevation model captures every boulder, path cut, and rock step. In a block build, those become jagged single-block spikes that make the terrain look like bad pixel art, not a mountain.
Third, the cost is not linear. Building and editing a 2.4-billion-column world takes time, RAM, and disk that a 25 m build doesn't need. Most people who abandon these projects don't run out of data. They run out of patience.
That said, there's a real case for fine detail. If you're building a single feature, like the Oeschinensee basin or a stretch of the Aletsch glacier margin, a 1 m or 2 m grid is worth the cost because you can see the difference at that scope. The rule that holds up is this: match the grid to the smallest area you'll actually look at closely, not to the whole country.
Coordinates, Heights, and Other Things That Break Builds
Most failed terrain imports don't fail because of the data. They fail because of the reference system. Swiss national mapping uses the CH1903+ / LV95 coordinate system, and heights are referenced to a national height datum. If your import tool assumes WGS 84 latitude and longitude, your mountains will land in the wrong place, sometimes kilometres away, and the error won't announce itself.
Here's the rule to follow: confirm the coordinate system of the file, then confirm your tool's expected projection before you import. If the two don't match, transform the data first. Don't trust the default.
Height datums cause a quieter problem. A few metres of offset won't ruin a 25 m build, but at 1:1 scale, a vertical error of two metres is two blocks of cliff where you expected a slope. Check whether the source heights and your tool's zero point use the same reference.
Glaciers add another trap. The Aletsch Glacier has been retreating for decades, and the surface you download is a snapshot. If you're building from a dataset that predates recent ice loss, your glacier will sit higher and wider than the real one. That's fine for a model, but label it honestly if you share it.
Licensing and Attribution
The source credits the map to swisstopo, and you should read the terms of use before you publish a build, a video, or a print. Open data doesn't mean unrestricted data in every case. Attribution requirements and permitted uses can differ between map products, elevation products, and imagery. Check the current swisstopo terms yourself, because I can't verify their exact wording from the source page.
A simple habit helps: keep a text file in each project folder that records the dataset name, the download date, the source URL, and the licence you checked. When someone asks where the mountain came from in two years, you'll have an answer.
Choosing the Right Elevation Source
Switzerland publishes several elevation products, and they're not interchangeable. swissALTI3D is the high-resolution model, with a grid spacing around 0.5 m. DHM25 is the coarser national model at 25 m. The gap between them is the gap between the 1 m and 25 m rows in the table above.
Here's where most people go wrong: they pick the high-resolution product because it sounds better, then discover their tool can't handle it. Start with DHM25 for a full-area overview. Once you know the build works, move to a finer product for the one valley or peak you want to show in detail. Build in layers, not all at once.
That said, don't treat the coarse model as a shortcut that's always good enough. For very steep faces, such as the north walls below the Eiger and Mönch, a 25 m grid smooths the relief so much that the face disappears. For those sections, the finer model is the only honest choice.
FAQ
What is the Switzerland 3D terrain model?
The Switzerland 3D terrain model is a free download of swisstopo map and elevation data for the Jungfrau-Aletsch UNESCO area, about 2,400 km². It's packaged as a 1.1 GB ZIP formatted for Prism, and it covers features such as the Eiger, Mönch, Jungfrau, the Aletsch glacier, and the Oeschinensee. It's a terrain dataset, not a finished 3D model.
How do I get swisstopo terrain data for free?
Download the ZIP from blocks.manifold3d.ch, which requires no registration. Then check the swisstopo terms of use for your intended use, and keep a record of the source and download date. Before importing, confirm the coordinate system and height reference, since a mismatch will misplace the terrain.
Why does a 1:1 block build get so large?
Each block represents one real metre at 1:1, so the number of blocks grows with the area. Across 2,400 km², the surface alone needs about 2.4 billion columns at 1 m spacing. Most builds work better at 10 m or 25 m for the whole area, with finer detail added to smaller sections.
What resolution should I use for a block terrain build?
Match the grid to the smallest area you'll examine closely. For full-area overviews, 25 m is practical. For one valley or glacier basin, 1 m or 2 m is worth the cost. Finer is not automatically better, because blocks are coarse by design and very fine grids produce jagged, noisy surfaces.
Why does my terrain appear in the wrong place?
The most likely cause is a coordinate system mismatch. Swiss national data uses CH1903+ / LV95, while many tools default to WGS 84. Transform the file to your tool's projection before importing, and check the height reference too, since a vertical offset of a few metres becomes several blocks at 1:1 scale.
Start Small and Measure Before You Scale
If you take one thing from this, make it this: build one valley at 1 m before you attempt the whole UNESCO area at any scale. Download the ZIP from blocks.manifold3d.ch, pick the Oeschinensee basin or one face of the Eiger, and time the import and the first render. Note the block count and your machine's memory use, then decide the grid for the full build from those numbers. Pass this to someone who's been staring at a crashed 1:1 import and wondering why their mountain is in the wrong country.