Stream Massive Point Clouds Over the Web

Use Cesium ion to tile your point clouds into 3D Tiles for efficient streaming and visualization with CesiumJS.

As the creators of 3D Tiles we leverage its full potential to ensure a lightweight and smooth streaming experience, even on mobile devices!

The Command Line Interface (CLI) is easy to integrate into any platform, including Linux, Windows, AWS, and Azure. It's built from the ground up for performance and optimized for the best visual experience and precision.

Dynamically Style with Per-Point Attributes

Cesium ion preserves per-point attributes so you can use them to control a point's color, size, or visibility using the 3D Tiles Styling Language.

You can dynamically style using a point's classification, intensity, or any custom expression combining multiple attributes. It's all done efficiently with GPU acceleration at run-time.

Faster Streaming with Point Compression

To optimize load times, Cesium ion provides an option to use Google's Draco point-cloud-aware compression library. The desired combination of precision and compression can easily be tweaked.

Compressing data at a finer precision than the original decreases file sizes at no loss. This highly detailed cabin was compressed from the original 4.48 GB LAS down to 0.64 GB.

In general, Cesium ion produces a 7x reduction compared to the source LAS files and a 3x reduction compared to LAZ. Decoding in CesiumJS is fast, using WebAssembly and done in parallel with web workers.

Gzip Draco
1.83 GB 0.67 GB
gzip Draco

Draco produces far better compression rates than general purposes algorithms like Gzip.

Easy to Use

Cesium ion brings together all the algorithms you need in an easy-to-use configurable CLI. Convert your raw LAS data to an efficient tileset that's ready for streaming in a single command.

High Performance

The point cloud pipeline is written in C++ and makes use of every available CPU core, giving it a significant performance advantage.

Its multithreaded architecture overlaps I/O and compute, and an out-of-core implementation allows it to scale to massive datasets.

The point cloud CLI can process around 5 million points per second on average hardware for in-core paths.

Efficient Tiling

Cesium ion produces tilesets that are optimized for the web. Tile sizes are tuned for efficient network traffic. Tiles are computed to be a visually representative subset at any given view using a precise geometric error so tilesets can stream faster.

The Hierarchical Level of Detail (HLOD) algorithm allows tiles to be loaded only when needed and contain just enough detail for any given view. Additive refinement allows higher resolution data to smoothly transition in.


Cesium ion performs a robust preprocessing step to detect the optimal configuration settings for your data. No need to waste time digging through layers of configuration.

We've tested it with everything from city-scale point clouds to detailed interior scenes, and optimized for both uniformly and non-uniformly distributed data sets.

Whether your data comes from LiDAR scanners on cars, satellites or anything in between, you can expect it to work with ion.

Millimeter Precision

Cesium is used by many to make real world decisions, and that's why precision is at the core of everything we do.

Cesium ion has no trouble with points at millimeter precision even on mobile devices. No data is lost at any step in the pipeline.

The leaf tiles maintain the highest resolution accuracy so once a view is fully refined you can measure and analyze with confidence.

Interested in trying it out?

Email Tim Rivenbark for a demo.

Technical Resources

Check out the full documentation for the on-premises 3D Tiling Pipeline.