Sentinel-2 super-resolution

A finer grid.
A clearer workflow.

Turn Sentinel-2 surface reflectance into a ten-band, georeferenced COG on a 2.5 m output grid. From your dashboard or API.

Create a free account Explore the demo

10 free credits (10 km²) after email verification. No card required.

1 credit = 1 km². €1 buys 4 credits.

Sentinel-2 input view of Urban retail blocks and housing · 640 m crop — Urban retail blocks and housing · 640 m crop

HyperSR v1 output view of Urban retail blocks and housing · 640 m crop — Urban retail blocks and housing · 640 m crop

train split · Sentinel-2 2023-05-24 · NEON reference 2023-05-26. Qualitative example; not held-out performance.

From scene to COG

Less setup.
More time with your data.

Choose a scene and your area. Review the quote. Download the output when processing completes.

01 / SELECT

Your area. Your scene.

Start with supported Sentinel-2 Level-2A imagery and a rectangular area of interest.

02 / CONFIRM

Know the cost first.

One credit covers one km² per scene. Review a server quote before starting each asynchronous job.

03 / DOWNLOAD

Keep the context.

Georeferenced imagery, quality masks and provenance, with private download links.

For developers: bands, API and Python

hypersr_v1 supports B02 · B03 · B04 · B05 · B06 · B07 · B08 · B8A · B11 · B12. SRGAN guidance and SEN2SR spectral fusion reconstruct detail; B10 and native 60 m bands are outside this model's contract.

Python

from hypersr import HyperSR

with HyperSR(api_key="YOUR_API_KEY") as client:
    quote = client.jobs.quote(
        scene_id="YOUR_SCENE_ID",
        bbox=(2.1, 41.3, 2.11, 41.31),
        model="hypersr_v1",
    )
    job = client.jobs.submit(quote_id=quote.id)
    result = client.jobs.wait(job.id)
    if result.status == "completed":
        client.jobs.download(result.id, "output.tif")

REST

POST /v1/jobs/quote
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "scene_id": "YOUR_SCENE_ID",
  "bbox": [2.1, 41.3, 2.11, 41.31],
  "model": "hypersr_v1",
  "output_format": "cog"
}

Examples require a configured endpoint and enabled model. Read the quickstart →

Reconstructed detail, with context

A 2.5 m output grid is not a native 2.5 m observation. Inspect dated comparisons and the 225-tile SEN2NEON benchmark before using the results.

Explore the evidence

A finer grid.A clearer workflow.

Less setup.More time with your data.