Industries

The decision your resolution unlocks.

One sourced use case, two candidates we can already support visually — and a backlog we won't illustrate until we can do it with real exports.

01 · Game & simulation terrain — the founding use case

Terrain of real places, without buying aerial surveys.

Build a plausible, coherent 1:1 terrain of any real place from free, current satellite passes — texture statistics drawn from real aerial surveys, layout faithful to the source.

Texture realism · KID-HP ↓ (benchmark)3 262 vs floor 18 297
Round-trip @10 m · pinned protocol0.898

Open the Live Map →

Bicubic upsample of the Sentinel-2 benchmark input — soft, blocky
SOURCEBENCHMARK INPUT
Generated 2.5 m texture of the same crop — field rows and buildings resolved
GENERATEDSEAN V5 · 2.5 M

02 · GIS base-map refresh — candidate

A current aerial look, between surveys.

National orthophoto cycles take years; Sentinel-2 passes every five days. On identical real inputs, our output leads the best public model by ≥1.7× on texture realism (4 631 vs 7 777).

See the evidence →

Sentinel-2 winter pass over Brežice, January 2024
SENTINEL-2PASS 2024-01-27
Sentinel-2 autumn pass over the same town, September 2024
SENTINEL-2PASS 2024-09-23
One town, one year, ten passes — the refresh rate is the product.

03 · Port & industrial monitoring — candidate

Readable situational imagery of active sites.

Containers, yards and berths at 2.5 m, refreshed with every pass. Industrial is one of the hardest strata — per-stratum KID-HP 4 340 (benchmark), and we publish that number rather than hide it.

Start a pilot conversation →

Sentinel-2 pass over the Port of Koper at 10 m
SENTINEL-2IN · 10 M
Generated 2.5 m image of the Port of Koper — terminals and rail yard visible
GENERATEDSEAN V5 · 2.5 M

Backlog — candidates without visuals yet

Digital twins & urban planning · environmental and land-use monitoring · film/VFX terrain backdrops. Each ships a spread only when we can support it with real exports — until then: talk to us.

Licensing: commercial framing applies to the 2.5 m stage (trained in-house). The 0.5 m research stage runs on non-commercial research weights and is not offered commercially.