How to View Point Clouds in the Browser (No Install)

The classic way to look at a point cloud is heavy: a licensed desktop package, a workstation to run it on, and tens of gigabytes to download first. Once the data has been converted to a browser-friendly format, though, LAS/LAZ opens with no installation at all — a colleague clicks a link and the data is on screen in seconds.

Upload and conversion flow

Raw LAS/LAZ files are not rendered directly; they are first converted into a hierarchical structure called an octree. The octree splits the data into spatial boxes and levels of detail, so the browser downloads only the region currently on screen at an appropriate resolution — not all 400 million points.

There are two routes:

1. Convert in the cloud — upload your LAS/LAZ and the conversion runs server-side. The per-file limit is 5 GB / 150 million points; it consumes processing tokens and requires the Professional plan (locked on the Free plan).
2. Upload a ready octree — run the conversion on your own machine (with CartaX Studio) and upload the output. This route is token-free and available on every plan. A ready octree has no separate size cap, only your storage quota applies — CartaX Studio is the path for datasets above the 150-million-point limit (no raw file size limit).

Current limits are listed in full on Supported Formats and Limits.

Navigating the viewer (orbit/zoom)

Controls follow the same logic as CAD and GIS packages: in the default Orbit mode one mouse button rotates and the other pans; the wheel zooms. The viewer's hint bar shows the current button assignments. The toolbar switches between Orbit / Fly / Earth navigation modes; double-clicking a point focuses the camera there.

Two settings do most of the work in making a cloud readable:

• Point Size — increase it on sparse data, decrease it on dense data
• EDL (eye-dome lighting) — shades depth differences between neighbouring points. On colourless or single-colour data it is the single most effective switch for revealing geometry.

There are seven background presets: Dark, OLED, Light, Gradient, Diagonal Night, Dark Forest and Warm Ground; on light-toned ground data a dark background raises contrast noticeably. The colour tuning section offers Gamma / Brightness / Contrast sliders and an automatic range for intensity mode. Take care when raising the Point Budget: a budget above 10 million points strains the GPU and RAM, and the viewer warns at that threshold.

Class filter and elevation colouring

The same data can be read through four colour modes:

Colour mode	What it shows	When to use it
RGB	Real colour from imagery	Presentation, overview
Elevation	Colour ramp by height (adjustable range)	Topography, reading slope
Intensity	Laser return strength	Surface material, road markings
Classification	LAS class code colour	Separating ground / vegetation / buildings

If the data is classified, the Classification panel lets you hide and show classes individually. Switching vegetation off exposes the ground and structures underneath — the fastest way to see on screen what you could not see on site. Files without class information show "No classification data", and points carrying no class code are grouped under Unclassified.

Cross-section (profile) tools

The Profile tool draws a line across the cloud and opens a 2D cross-section beneath it. A section width setting controls how many metres either side of the line contribute points — a narrow band gives a crisp section, a wide band a more statistical view. Inside the section panel you drag to pan, scroll to zoom and double-click to fit; moving the cursor along the band tracks the corresponding position in the 3D view.

Clip view isolates the volume you care about: draw a box (Box Clip) or a free shape (Polygon Clip) and hide either the inside or the outside. A clip you like can be saved to the project as a named Clip View and returned to later.

Selections can also be exported: the output arrives as a single ZIP containing CSV + LAS + 2D/3D DXF. So you can select in the browser and continue in your desktop software.

The share (guest) viewer

A guest opening a point cloud you shared by link gets a read-only viewer: the right panel (file info, colour mode, point settings) plus measurement and note tools; Profile, Clip and ZIP export are not offered to guests. Guest measurements are temporary and are not saved to the project. Analysis and point-cloud share links require a desktop browser; they do not open on mobile devices.

Performance tips

• Point Budget is how many points are held on screen at once. If the view stutters, this is the first setting to lower; raising it increases density, not detail.
• Do not widen the FOV unnecessarily — a wide field of view loads more octree nodes simultaneously.
• Use clipping: hiding everything outside a box reduces not just what you see but how many nodes get downloaded.
• Zoom into the area you care about instead of waiting on a distant overview — the octree fetches detail as you approach; waiting from afar downloads nothing.
• On very large datasets, splitting the data into several layers stays smoother than one enormous one.

Make the browser use your discrete graphics card

On laptops with both an integrated (Intel/AMD) and a discrete (NVIDIA/AMD) GPU, Windows and Chrome/Edge usually run the browser on the integrated GPU to save battery. CartaX tells the browser it prefers the powerful GPU, but on Windows the final decision belongs to the operating system; no web page can override it on its own. You can see which GPU is in use on chrome://gpu (edge://gpu in Edge) — look for the row marked *ACTIVE*. If it shows the integrated GPU, any one of these is enough:

1. Windows Settings → System → Display → Graphics → add your browser to the list → choose High performance (persistent; the recommended route).
2. Enable the chrome://flags/#force-high-performance-gpu flag and restart the browser.
3. NVIDIA Control Panel → Manage 3D Settings → Program Settings → High-performance NVIDIA processor for your browser.

The same setting also speeds up the 3D map (terrain and 3D Tiles) alongside the point cloud; desktop PCs with a single graphics card do not need this step.

Summary

Question	Answer
Any installation?	No — a browser is enough, for recipients too
Can raw LAS/LAZ open directly?	No; it is converted to an octree first
Cloud conversion limit?	5 GB / 150 million points (uses tokens, Professional plan)
Anything larger?	Ready octree via CartaX Studio — token-free, storage quota only, no raw file size limit
Colour modes?	RGB, Elevation, Intensity, Classification
Cross-sections?	Profile tool with adjustable section width
Export?	ZIP: CSV + LAS + 2D/3D DXF