Learn

What Is a Tile Pyramid?

A tile pyramid is a multiscale hierarchy of map tiles in which each higher zoom level divides the map into more tiles and can carry more geographic detail.

Geobble
introductoryexplainerWeb Maps Tiles & Performance

Explain the multiscale logic of tile pyramids, tile counts, level of detail, generalisation, and why high zoom is not simply a larger version of low zoom.

What Is a Tile Pyramid?

A tile pyramid is a hierarchy of map tiles at several zoom levels. Each level divides the mapped world into more tiles, allowing a client to request a progressively smaller geographic area with more detail as the user zooms in.

In the common web-map model, zoom 0 contains one world tile, zoom 1 contains a 2 × 2 grid, zoom 2 contains 4 × 4, and so on.

At zoom z, the grid has 2^z tile positions across each dimension.

Why the pyramid exists

A world map and a street map do not need the same level of detail: at global scale, tiny roads, building footprints, and every vertex of a detailed coastline would create visual clutter and unnecessary data transfer, while at street scale those details can become essential.

A tile pyramid lets the publication maintain several levels of representation rather than forcing one geometry density into every scale.

Higher zoom means more tiles, not simply bigger tiles

The tiles themselves normally stay at a fixed pixel or coordinate extent, while the geographic area represented by each tile becomes smaller as zoom increases.

Because each tile has four children in the next quadtree level, the number of possible tile positions grows rapidly:

zoom 0: 1 tile
zoom 1: 4 tiles
zoom 2: 16 tiles
zoom 3: 64 tiles
...

A full global pyramid to very high zoom would contain an enormous number of possible tiles. Real tilesets usually cover only their relevant geography and zoom range, and may generate tiles on demand or package them in archives.

Detail should change with zoom

A good tile pyramid is not merely the full-resolution source repeated at every zoom.

At lower zooms, publication data may be:

  • simplified;

  • filtered;

  • aggregated;

  • represented with fewer attributes;

  • replaced with a different geometry representation.

For example, individual buildings may disappear at regional scale. Minor roads may appear only after a threshold. Millions of points may become clusters or counts.

This is cartographic generalisation as much as performance optimisation.

Zoom level is not a universal physical scale

Web-map zoom is an integer level in a particular tile scheme. Physical scale also depends on latitude, screen pixel density, viewport, and projection.

A map at zoom 10 does not have one universal paper-style scale everywhere on Earth.

Map Scale vs Zoom Level explains that distinction.

Pyramid design affects storage and quality

Choosing a maximum zoom too high can produce huge tile archives and imply precision the source data do not support. Choosing it too low, however, can make users zoom past the last useful representation and encounter blurry raster resampling, missing vector detail, or an abruptly empty map.

The useful maximum zoom follows source resolution and user need, not simply the maximum value supported by a library.

Tile pyramids can contain raster or vector tiles

Raster pyramids store rendered images at each scale, whereas vector pyramids store encoded vector features that a client styles.

The multiscale indexing idea is shared even though the tile contents differ.

References

  1. OpenStreetMap Slippy Map Tilenames. Documents the conventional quadtree tile grid and zoom-level addressing.

Related content