Point clouds used to require specialist desktop software to view. CloudCompare, QGIS, Leica Cyclone — all requiring installation, setup, and a learning curve that most non-specialist clients aren’t willing to tackle.
That’s no longer the case. Point clouds can be viewed in any modern browser, in real time, with no software installation — using WebGL-based streaming viewers.
Here’s how it works and how to use it.
The technology: Potree and WebGL
The key technology enabling browser-based point cloud viewing is Potree — a WebGL renderer developed at TU Wien that uses an octree streaming architecture to load only the portion of the point cloud visible at the current zoom level.
This is important: a 5 GB LAZ file is not loaded entirely into the browser. Instead, a preprocessed octree is served from a web server, and the browser requests tiles on demand as you navigate. The viewer feels responsive because you’re never loading more than what’s visible.
The renderer uses WebGL (Web Graphics Library) — the same browser API used by browser-based games and 3D web applications. Modern browsers (Chrome, Firefox, Safari, Edge) all support WebGL by default.
What you need to view a point cloud in a browser
Two things:
- The point cloud data, preprocessed into the Potree streaming format
- A web server to serve the tiles, and a Potree HTML viewer page
The preprocessing step is where the friction currently exists. Raw LAS/LAZ files can’t be loaded directly into a browser — they need to be converted to the Potree octree format first, using PotreeConverter (a command-line tool).
Option A: Self-hosted Potree
- Download and run PotreeConverter on your LAZ file:
PotreeConverter input.las -o output_folder/ - Host the output folder on a web server (HTTPS required)
- Access the viewer via the generated HTML page
This works, but requires server infrastructure and technical knowledge.
Option B: Use a platform that handles it for you
Upload your LAZ/LAS file to a platform like Swyvl. The platform runs PotreeConverter (or equivalent), hosts the tiles, and gives you a URL that works in any browser.
No server to manage. No PotreeConverter to install. Upload the file → get a shareable link → done.
What point cloud viewing looks like in a browser
A modern browser-based point cloud viewer (via Potree) provides:
Navigation:
- Left click + drag: Rotate around the point cloud
- Right click + drag: Pan
- Scroll wheel: Zoom in/out
Rendering:
- Point size adjustment (smaller = more detail visible at full zoom; larger = better overview)
- Colour mode: RGB (actual colour), intensity (signal strength), elevation (height-based colour gradient), classification (colour by class)
- EDL (Eye Dome Lighting): A shading technique that adds depth cues without a traditional lighting model — makes point clouds dramatically easier to read for non-technical viewers
Tools (vary by implementation):
- Distance measurement: Click two points to measure distance
- Area measurement: Define a polygon and measure area
- Volume measurement: Define a region and estimate volume
- Cross-section / clipping: Clip a region to see interior surfaces
- Profile: Draw a line and see a 2D cross-section along that line
For engineering and survey use, the measurement tools are important — clients can take their own measurements from the delivered point cloud without needing to come back to you.
Supported formats
Most browser-based point cloud viewers work with data converted from:
- LAS / LAZ (most common — airborne LiDAR, photogrammetry)
- E57 (terrestrial scanners — Leica, Faro, Trimble)
- XYZ / CSV (generic point formats — convert to LAS first for best results)
- PLY (some viewers support PLY directly)
Performance considerations
Browser-based point cloud viewing is GPU-intensive. Performance varies significantly:
| Setup | Performance |
|---|---|
| Modern laptop with discrete GPU | Excellent — handles 100M+ point datasets |
| Modern laptop with integrated GPU | Good — handles 10-50M point datasets well |
| Older laptop with integrated GPU | Fair — smaller datasets fine, large datasets may lag |
| Mobile (phone/tablet) | Variable — small datasets work; large ones may struggle |
| Desktop workstation | Excellent |
For the best client experience, test on a device similar to what your client will use. For clients known to be viewing on lower-spec hardware, consider whether you can deliver a lighter-weight representation (less dense, classified ground only, etc.).
Coordinate systems and georeferencing
A point cloud that’s been correctly georeferenced will appear in the right location when overlaid with a basemap in a viewer that supports it. Potree-based viewers can display a Bing Maps or OpenStreetMap basemap underneath the point cloud — useful context for clients to understand the survey area.
Ensure your point cloud is in a projected coordinate system (not geographic lat/lon) before conversion. Common choices:
- GDA2020 MGA zones (Australia)
- UTM zones (international)
- State plane (US)
For clients: what they need to do
Nothing. That’s the point.
A client who receives a Swyvl link to a point cloud:
- Clicks the link
- Sees the point cloud loading in their browser (takes 5-15 seconds for a large dataset)
- Uses left/right click and scroll to navigate — intuitive within a minute
No software download. No account required. No GIS knowledge needed.
Browser-based point cloud viewing is mature technology that’s been available since roughly 2016. The barrier has never been technical — it’s been the infrastructure required to convert, host, and serve the data. That infrastructure is now available as a service. There’s no longer a good reason to deliver point clouds as ZIP files.