We are pleased to announce the release of Virtual Geo version 9.3, a major update delivered in October 2025, bringing together significant performance improvements, extended format support, and new rendering and simulation capabilities.
Key highlights of Virtual Geo 9.3 :
Space & Simulation Capabilities


Multiresolution H3 Tiling and Culling Support

Improved Performance & Rendering Quality
Extended Geospatial Format Support
Networking & Data Access
25 years ago, two years before our American colleagues at Keyhole, our pioneering teams created one of the very first globe viewer, VirtualGeo, for the French Space Agency.
It is today the Digital Twin Earth system at the heart of many critical systems and C2s for civil protection, defense, space, and industrial applications and products such as Crimson, Boreades, Starlinx, directCGF.
This birthday is also an opportunity for us to remember that our journey wouldn't have been possible without the support of the European Commission through the Virtual-Planet and Virtual-City projects, which have enabled to turn VirtualGeo into the cutting-edge, widely adopted product it is today.
Time is flying!
![]() |
![]() |
The latest release of VirtualGeo brings you many enhancements and technical innovations:
Please check the changelogs of our components to find all the fixes and improvements !
VirtualGeo 9.0 introduces a lot of changes internally in the engine to push forward with modern technology and brings two exciting new features: a new renderer which supports modern graphics back-end such as DirectX 12, Vulkan or Metal, and a plugin for Unreal Engine to integrate VirtualGeo technology into an Unreal Engine project! More news and information about the plugin will comes very soon.
To enable these new features, VirtualGeo SDK is now split between a core which manages import of various formats, data management (including LOD selection from a viewport), and query and computation on imported data and export various formats, and a rendering module which manages visualization, interactive tools to create and edit data, etc.
The legacy renderer based on OpenGL is still available for compatibility reasons, because some less-used features have been deprecated and are not available in the new renderer.
With the new renderer, a huge refactor has been done on the internals of the engine which brings impressive improvements for performance.
On our different tests, we measure from 50% to 700% improvements depending on the context and the data (from terrain rendering to complex 3D Mesh, or just displaying many dynamic elements with complex decorations).
| Tests | 8.1 (µs) | 9.1 (µs) | Factor |
| Performance Terrain | 768 | 245 | 3,13 |
| Performance Terrain 3 | 2838 | 2144 | 1,32 |
| Performance Terrain Intervisibility | 1759 | 572 | 3,08 |
| Performance Surfaces | 14454 | 2038 | 7,09 |
| Performance Instancing | 3206 | 2357 | 1,36 |
| Performance Declutter | 62403 | 33861 | 1,84 |
| Performance Terrain + Mesh | 3437 | 1572 | 2,19 |
| Many Entities Performance | 13717 | 3201 | 4,29 |
At CSGROUP, we consider OGC 3DTiles as an important standard for the 3D cartographic world.
In this new version, we can now import 3D-Tiles with GLTF 2.0 support or instancing data to bring our support closer to 3DTiles Next specification.
Also, the improved support allows us to now integrate 3D data directly from Google Maps ! Google has opened its 3D data, based on the 3D Tiles format. VirtualGeo 9.0 now supports loading directly these data into your application. Please check Google page to find more information on how to include Google data.
VirtualGeo is available to develop a Web application thanks to WebAssembly technology which enables to integrate code in native language such as VirtualGeo and enables to brings high performance rendering to the Web.
For version 9.0, VirtualGeo now supports using multi-thread on its WebAssembly version to brings performance closer to native version.
Multi-thread is supported on major browsers (Chrome >= 74, Firefox >= 79, Safari >= 14.1) but need for security concerns must be explicitly activated on the Web page.
With VirtualGeo Server it is now possible to dynamically import and optimise data on the fly coming from data collected in real-time on the field. For example, data can include real-time lidar point cloud from robot scanning an are.
The real-time data is dynamically updated in VirtualGeo clients, taking into account data updates on the server, while still minimizing data transfer according to changes. VirtualGeo Server use Web Socket internally to provides real-time and smart update.
For the next version of our #VirtualGeo software suite coming out in May 2023, we are pleased to announce the first official release of the Unreal Engine plugin ! With this plugin, users can easily import, visualise and analyse any #3d #geospatialdata to create real-time and accurate game-standard graphic rendering for simulations.
VirtualGeo will still support the Certigo sovereign rendering engine for embedded and more critical applications.
Given that VirtualGeo supports many standard geospatial data formats and protocols such as OGC CDB, OGC 3D Tiles, OpenFlight and streaming protocols (WMS, WMTS, WFS), these can be easily imported into Unreal and displayed in real time. No need for complex setups, no pre-processing constraints... and the level of performance is simply unprecedented! The terrain built with VirtualGeo supports complex lighting and physics, it is completely integrated into Unreal Engine to construct advanced simulations.
Please note this first release is fully compatible with the latest version of Unreal Engine (v 5.1) to fully support world-wide rendering based on the Large World Coordinates (LWC) in Unreal Engine 5. Compatibility with the 4.27 version is also available on demand.
Check out this video, which demonstrates the VirtualGeo Unreal plugin.
Cesium is a well-known competitive solution of VirtualGeo Web, designed to display high-resolution mapping data in a Web page.
Both offer a similar set of Web features, but looking “under the hood”, the approach is very different.
VirtualGeo is based on a native engine developed in C++, which is translated into Web Assembly, see here for more details.
CesiumJS is based on a pure Javascript engine. Thanks to our 3DTIles support, it is now quite easy to compare both approaches and engines in terms of memory and CPU consumption on a same dataset.
As the base dataset for our comparison, we have chosen to use the highly detailed city of Strasbourg
See the result below :
