Softwr

Maps · head to head

Fused vs Potree

Fused logo

Fused

Maps

Serverless Python for geospatial: write a UDF, get a map tile endpoint and an HTTP API

From
Free
Rated
-
Full reviewWebsite
Potree logo

Potree

Maps

Open source WebGL viewer for publishing very large point clouds in a browser

From
Free
Rated
-
Full reviewWebsite

The short version

  • Each has a real cost: Fused the price steps are brutal: $20 to $200 to $2,000 a month with no intermediate tier, so a two-person team that outgrows Starter faces a tenfold increase for what is often a modest capacity gap.; Potree there is no vendor, no support contract and no service level, so anything you build on Potree is your organisation’s responsibility to maintain when a browser update breaks rendering.
  • They diverge on capability: Fused covers User defined functions, Potree covers Octree streaming.
  • Prices and features above were last checked on 1 September 2026.

Where they differ

Only the attributes on which Fused and Potree actually diverge.

Attributes where Fused and Potree differ
AttributeFusedPotree
Pricing modelPer month by compute tierOpen source, no licence fee
PlatformsWeb, Linux, macOS, WindowsWeb, Windows, macOS, Linux

Identical on both: starting price (Free), free tier (Yes), user rating (Not yet rated), category (Maps).

What each one covers

Drawn from each product's published feature list. An absence here means we hold no record of it - not that the product lacks it.

Only in Fused

  • User defined functions
  • Instant tile endpoints
  • HTTP API
  • Caching
  • Workbench
  • Data stack connectors
  • Public UDF catalogue
  • Local execution

Only in Potree

  • Octree streaming
  • Large dataset rendering
  • Measurement tools
  • Clipping and profiles
  • Annotations
  • Attribute colouring
  • PotreeDesktop
  • Self-hosting

What people use each for

The jobs each tool is most often brought in to do.

Fused

  • A data team that wants to publish a dynamic layer computed from a large Parquet dataset without standing up and operating a tile servernot Potree
  • An analyst prototyping against Overture, Sentinel or Census data who needs a working map in an afternoon rather than a data engineering ticketnot Potree
  • An application that needs a geospatial computation exposed as an HTTP endpoint, where the alternative is a container, a queue and an on-call rotanot Potree
  • A consultancy delivering an interactive constraint or suitability map to a client, computed live from source data rather than pre-baked into static tilesnot Potree

Potree

  • A drone survey firm delivering a point cloud to a client who has no desktop software, by publishing a browser viewer link instead of a 40 GB LAZ filenot Fused
  • A heritage or archaeology project publishing a scanned site for public access without paying per-gigabyte hosting to a commercial platformnot Fused
  • An engineering consultancy embedding a point cloud viewer inside its own client portal without licensing a commercial web viewernot Fused
  • A university or research group that needs measurement and profiling in a shareable viewer for a dataset too large to emailnot Fused

Where each one falls short

Documented limitations, not opinions. Every one is a constraint you would hit in normal use.

Fused

  • The price steps are brutal: $20 to $200 to $2,000 a month with no intermediate tier, so a two-person team that outgrows Starter faces a tenfold increase for what is often a modest capacity gap.
  • Published tiers do not state the actual compute, storage or request allowances, so the only way to know whether $200 covers your workload is to run it and find out.
  • Fused is at pre-seed stage with a small team, and putting a customer-facing map behind a startup’s serverless platform is a continuity risk that a procurement review will flag.
  • It is Python-only and assumes competence with cloud-native geospatial formats; there is no path for an analyst who works in a desktop GIS and not in code.
  • Serverless execution means cold starts and per-tile latency on uncached views, so a first load of a heavy UDF over a wide area is noticeably slower than a pre-rendered tile set.

Potree

  • There is no vendor, no support contract and no service level, so anything you build on Potree is your organisation’s responsibility to maintain when a browser update breaks rendering.
  • Conversion is a separate manual step with PotreeConverter, and building an automated publishing pipeline around it is engineering work the project does not provide.
  • Rendering quality and performance depend on the client machine and WebGL support, so a client on a locked-down corporate laptop with poor graphics may see a viewer that stutters or fails outright.
  • A WebGPU rewrite, Potree-Next, is in progress and is not a drop-in upgrade, so current integrations face an eventual migration with no vendor to manage it.
  • There is no access control, user management or audit layer; anyone with the URL sees the data unless you build authentication around the static files yourself, which is a real problem for confidential sites.

Pricing, plan by plan

Fused

Free
  • FreeFree
    • Workbench access
    • AI assistant
    • Local UDF runs
  • Starter$20/month
    • Secrets management
    • Higher compute limits
    • Office hours support
  • Developer$200/month
    • Dedicated jobs
    • GitHub integration
    • Expanded compute, storage and AI token allowance
  • Professional$2000/month
    • Team collaboration
    • Dedicated training
    • Substantially increased usage limits

Potree

Free
  • PotreeFree
    • BSD-style open source licence
    • Unlimited datasets and viewers
    • Self-hosted on any static web server

Which should you pick?

Choose Fused if

  • You need user defined functions.
  • You want to start without paying.
  • You work on Web, Linux, macOS, Windows.
  • You also want instant tile endpoints.
Fused review

Choose Potree if

  • You need octree streaming.
  • You want to start without paying.
  • You work on Web, Windows, macOS, Linux.
  • You also want large dataset rendering.
Potree review

Questions people ask

Is Fused or Potree better?
Neither clearly leads. Fused starts at Free and Potree at Free, and user ratings are close enough to be indistinguishable. Choose on capability and platform support.
Which is cheaper, Fused or Potree?
Fused starts at Free and Potree at Free.
Does Fused or Potree run on more platforms?
Fused runs on Web, Linux, macOS, Windows. Potree runs on Web, Windows, macOS, Linux.
Can I use Fused for free?
Both have a free tier, so you can try either at no cost before committing.
What is Fused best used for?
Fused is most often used for a data team that wants to publish a dynamic layer computed from a large parquet dataset without standing up and operating a tile server, an analyst prototyping against overture, sentinel or census data who needs a working map in an afternoon rather than a data engineering ticket, an application that needs a geospatial computation exposed as an http endpoint, where the alternative is a container, a queue and an on-call rota, a consultancy delivering an interactive constraint or suitability map to a client, computed live from source data rather than pre-baked into static tiles. Of those, a data team that wants to publish a dynamic layer computed from a large parquet dataset without standing up and operating a tile server and an analyst prototyping against overture, sentinel or census data who needs a working map in an afternoon rather than a data engineering ticket are not what Potree is typically brought in for.
What can Fused do that Potree cannot?
Fused covers User defined functions, Instant tile endpoints, HTTP API, Caching. Potree covers Octree streaming, Large dataset rendering, Measurement tools, Clipping and profiles.

Answered from the vendors’ own pages

Fused: What does Fused actually give me?

A Python function becomes a map tile endpoint and an HTTP API on managed infrastructure, with caching, without you running a server.

Potree: Does Potree cost anything?

No. It is open source and free. You supply the hosting, which for a static viewer is inexpensive.

Fused: How much does it cost?

Free, then $20, $200 and $2,000 per month for Starter, Developer and Professional. Usage allowances are not itemised publicly.

Potree: Is there commercial support?

No. There is no vendor and no support contract. Support is GitHub issues and the community.

Fused: Can I run UDFs locally?

Yes, through the Python SDK, which is how most development is done before deploying.

Potree: Who maintains it?

It originated at TU Wien and is maintained principally by Markus Schütz, with community contributions.

Fused: Does it replace PostGIS or BigQuery?

No. It reads from them. Fused is the compute and serving layer, not the store.

Potree: Can it handle very large clouds?

Yes, hundreds of millions of points, because PotreeConverter builds a multi-resolution octree that streams only what is visible.

Fused: Is it a mature company?

No. It is a small pre-seed San Francisco company. The technology is real; the vendor risk is real too.

Potree: Is my data private?

Only if you make it so. Potree serves static files with no built-in authentication, so access control is entirely your responsibility.

Share

Related pages

More on Fused

More on Potree

Keep looking

Other head to heads