Reduce the points. Keep a separate result.

Choose a target count or minimum spacing, then download a smaller point cloud.

Point cloud preview

Bring your scan into view

Add a cloud, or try the sample to explore.

Rotate · Zoom · Inspect
Preview uses up to 80,000 points
Points—
File size—
Extent (X / Y / Z)—

How to subsample your point cloud

  1. Upload a point cloud and select it in Your clouds.
  2. Choose Target point count for random sampling, or Minimum spacing for spatial sampling.
  3. Run Subsample cloud. Select the new result to inspect, download or process it again.

Try an example

Try the sample pavilion and enter 10,000 as the target point count. Compare the point count before and after by selecting the original and the result in Your clouds.

Before you start

Subsampling runs on all source points, not just the preview. Random sampling does not guarantee that small features survive. Minimum spacing uses the source coordinate units; no unit conversion is performed.

Which sampling method should I choose?

Use a target count when you need a specific number of points. Use minimum spacing when you want to remove points that are too close together. Neither method reconstructs surfaces or fills gaps.

Files stay in a temporary 2-hour session. Switching tools in this tab keeps your uploaded clouds and results. Read about file privacy.

Choose a useful sampling setting

Start with the number of points you need

For a two-million-point source, a target of 200,000 keeps roughly one tenth of the points. This is an example, not a recommended setting for every scan. Choose a target below the source count, inspect the result and keep more points if small features become hard to distinguish.

Use spacing when coordinate scale matters

Minimum spacing is measured in the source units. If the coordinates are in meters, a spacing of 0.1 means 10 centimeters. Spatial sampling does not promise a particular output count; the count depends on the distribution of the source points.

Compare against the original

Each run creates a separate result. Select the original again before testing another setting, so you compare two reductions of the same source. Repeatedly sampling a reduced result cannot recover points already removed.

Point count or minimum spacing?
MethodChoose it whenTradeoff
Target point countYou have a point-count budgetRandom selection can remove small or sparsely sampled features
Minimum spacingYou want separation between retained pointsThe output count depends on spacing and the source distribution

Common questions

Does subsampling reduce file size by the same percentage?

Not necessarily. File size also depends on the format and retained attributes. A PLY result may still be larger than a compressed LAZ source even with fewer points.

Should I crop or subsample first?

If you only need one region, crop that region first. You can then spend your target point count on that region instead of on parts you will discard.

Service limits: 100 MB per upload, 5,000,000 points per cloud, one running job, 2-hour sessions. E57 merges up to 64 scans using stored poses, with common RGB/intensity only. Registration and cloud-to-cloud distance are not available.