Insights

3D Digital Twins: Stunning on Day One, Useless by Day Thirty

3D Digital Twins: Stunning on Day One, Useless by Day Thirty

Two stories from the past two weeks point the same way. Defense and intelligence teams want 3D environments that stay current, and the formats that make those environments photorealistic are moving into open, standards-based tools. For anyone fielding geospatial systems on a mission network, both trends raise the same practical question: where does that 3D data live, and can it run when the network can’t reach the cloud?

From one-off collections to a living baseline

A recent Federal News Network commentary argues that defense and intelligence digital twins should stop treating each collection as a standalone snapshot. Instead, the twin becomes a baseline that every new collection is measured against. The author describes a layered model. Remote sensing builds the physical foundation, persistent data shows change over time, analytics and GeoAI flag what matters, and integration with operational systems turns that understanding into action.

The piece is deliberately sensor-agnostic. Lidar supplies accurate terrain and structure, aerial and satellite imagery add visual context, SAR collects through darkness and weather, and thermal and spectral sensors reveal what conventional imagery misses. Existing GIS, engineering, BIM and CAD data fill in the rest. The author argues that no single sensor is the right answer, and the mission should decide the mix. The commentary is also clear about the purpose: “The goal isn’t to remove the analyst.” The twin should help analysts spend their time where their judgment counts.

Gaussian splats head into open-source GIS

On the software side, Radiance Fields reports that Lutra Consulting and North Road have launched a crowdfunding campaign of €85,000 to bring Gaussian splats to QGIS. The goal is to render splats in both 2D and 3D map views. Today QGIS displays only triangle meshes through its 3D Tiles and I3S support.

The format choice stands out. The splats would arrive as 3D Tiles, using an industry glTF extension for Gaussian splatting that several 3D vendors developed together. Work is planned for QGIS 4.6 through 4.10 across 2027, and it begins only if the campaign reaches its target before it closes in early December. The package also includes atmosphere effects, scene presets, keyframe animation with video export, and gamepad navigation for 3D scenes.

Why it matters for defense geospatial

Put the two stories together and a pattern emerges. The value of a 3D environment comes less from how good it looks on day one and more from how reliably it can be refreshed, compared and shared. A digital twin used as a change baseline needs a repeatable pipeline: collect, process into a standard format, publish, and compare against what came before. That pipeline breaks if each new data type requires a new viewer, a new license or a new cloud service.

Open formats are what keep the pipeline intact. When splats, meshes and point-cloud products all travel as 3D Tiles, a unit can process drone or lidar collection once and use it across tools without reformatting. Broader open-source support also lowers the cost for analysts who need to inspect 3D content without a dedicated workstation license.

There is also a deployment question that commercial coverage often skips. Many defense users work on restricted or fully disconnected networks. A 3D baseline is only useful if it sits on infrastructure the mission controls and keeps streaming when the uplink is gone. Interoperability and self-hosting are core requirements for operational 3D, not extras.

There are limits too. Photorealistic formats such as splats render appearance very well. Measurement-heavy work like line of sight and trafficability still depends on accurate terrain and geometry underneath. A sound 3D baseline carries both: accurate geometry for analysis, and realistic texture for recognition and planning.

What to watch

The QGIS campaign closes in early December, and whether it reaches its €85,000 target will show how much demand there is for splats in open-source desktop GIS. Over the next year, watch whether the shared glTF splat extension becomes the default way to deliver splats across commercial and open-source viewers, and whether defense programs begin writing 3D Tiles delivery into collection requirements. Also watch whether digital twin efforts are judged on how often their baselines are refreshed and compared, not just on how detailed the first model looks. That shift would move 3D from a showcase product to everyday analytic infrastructure.

Need 3D that keeps streaming when the uplink is gone?