LiDAR point clouds, as a mesh you can use.

Drop a LAS or LAZ file and Studio reads it in your browser: keep the ground for a clean terrain model, or keep everything for a surface model with buildings and trees. You choose the cell size, the result is a mesh, and nothing is uploaded to a processing queue.

Explore Mixos

A LiDAR surface model with buildings and trees on the left and bare-earth ground on the right — Boulder, Colorado: one real USGS 3DEP tile, 1 m cells. Live Studio viewport, not retouched.

LAS and LAZ

Compressed LAZ and plain LAS both open, decoded in a background worker so the page stays responsive.

Ground or everything

Keep only points classified as ground for a bare-earth terrain, or keep every class for a surface model that includes buildings and canopy.

You pick the detail

Set the grid cell size to trade detail for speed; holes under buildings and water are filled, and flat areas are simplified.

Colour where the scan has it

Points with RGB colour carry it into the mesh; otherwise you can colour by height or classification before you assign materials.

From point cloud to mesh

Studio's LiDAR import options showing point count, classes and cell size — The import sheet for a real 16-million-point Boulder, Colorado tile, read from its header before decoding.

See the tile before you commit

The header is read first: point count, extents and the classes in the file, with a cap and subsampling for big tiles.

A mesh gridded from all LiDAR classes, with buildings and trees — Boulder, Colorado, USGS 3DEP tile: surface model at 1 m cells, every class except noise, coloured by height.

Everything: the surface model

Keep every class for roofs, canopy and ground together.

Bare-earth terrain gridded from ground-classified LiDAR points — The same tile, camera and light: ground class only at 1 m cells, with gaps under buildings and trees filled.

Ground only: bare earth

Keep the ground class and the gaps under trees and buildings are filled.

Why Mixos

Why LiDAR, and why a mesh

Global elevation data is about 30 m per sample in most of the world. Airborne LiDAR is often under 1 m, which is the difference between a hill and the hill's footpaths, kerbs, banks and walls. Survey agencies in many countries publish it free — in the US, the USGS 3D Elevation Program covers most of the country — usually as LAS or compressed LAZ tiles.

A point cloud is not something most 3D tools can render or print. It needs gridding into a surface, with a decision about which points count. Ground-only gives the bare earth beneath trees and buildings, which is what a site model, a print or a flood study wants. Every class gives the surface a drone would see, roofs and canopy included.

Studio does that conversion in the browser. It reads the header first, so you see the point count, extents and available classes before committing, then grids the points you chose at the cell size you chose, fills the gaps and simplifies flat ground. The result is an ordinary mesh, recentred near the origin in metres with the original offset kept in its metadata.

Large tiles are large. A single LiDAR tile can hold tens of millions of points; the import offers a point cap and subsampling so the scene stays workable, and tells you what it kept.

Where to get LiDAR

USGS 3D Elevation Program

Free LiDAR and elevation products for the United States, downloadable as LAS/LAZ tiles.

Frequently asked questions

How do I convert a LAS or LAZ file to a mesh?

Import it into Studio, choose ground-only or all classes and a cell size, and it becomes a terrain or surface mesh. Importing and editing are free; exporting the mesh is the paid step.

Is my point cloud uploaded anywhere?

No. The file is decoded and gridded in your browser.

Can I remove trees and buildings?

Yes. Ground-only keeps the points classified as ground and fills the gaps beneath trees and buildings, giving bare-earth terrain.

How big a file can it handle?

It depends on your device's memory. Large tiles can be capped or subsampled on import, and the import reports how many points it actually used.