What does Umbra SceneStream do?

Umbra SceneStream generates a hierarchical and spatial reconstruction of a given input 3D dataset. The generated output is a set of spatially divided 3D meshes stored in a hierarchical Level of Detail (LOD) tree structure.

What does it enable?


Guaranteed rendering performance without perceptible loss of quality


Real-time rendering of arbitrarily complex 3D at its native resolution on a wide range of target platforms


Efficient delivery as the output data format is tailored for streaming


Fast load-up times as assets don’t need to be fully loaded to start rendering


Fire-and-forget workflow for the optimization


Fast turn-around times from asset creation to rendering using distributed infrastructure

Cloud platform

Optimizing, hosting, and delivering complex 3D scenes can be an immensely resource-heavy operation. To tackle that, we developed Umbra SceneStream's processing pipeline with a distributed infrastructure. It allows perfectly parallel operations and produces an output ideal for high real-time rendering performance.

Fast Optimization

Each volume in the generated structure is a computation job for a single CPU core, resulting in any number of optimization jobs to be processed in parallel.

Content availability

Any end client can access the 3D content on-demand, resulting in no need to unload full datasets to each end client. Delivery is possible from a local drive or over the network depending on requirements.

Progressive delivery

Applications can fetch only the data that is needed per a given camera viewport instead of loading the full scene before application startup.

Get Early Access

We’re currently looking for development partners to grant early access.

Contact Us



Need help with getting started or isn't something working like it's supposed to? Just contact our support team and we would love to help you out.

Our offices

Helsinki, Finland

Umbra Headquarters
Kalevankatu 30
00100, Helsinki