Features
Point clouds
How points are drawn and how large they are
Point cloud tiles (pnts, or glTF tiles made of points) stream like meshes. Their settings are in
the Nodes: Geometry panel: how each point is drawn, and how large.
Display
| Mode | What you get |
|---|---|
| Point cloud | Blender points. The cheapest. Colours show in Material Preview and Rendered; Solid shows the material colour. |
| Tiny tetrahedra | Four vertices per point, as a real mesh. Colours also show in Solid, with Color > Attribute in the viewport shading popover. |
| Disc splats | A small disc on every point, turned along the point's normal when the tile has normals, otherwise toward the view. |
Radius
| Mode | Point size |
|---|---|
| Attenuated (default) | From the spacing of the points in each tile, times Point scale, kept between Min px and Max px on screen. Coarse tiles get large points, fine tiles small ones, so the cloud looks filled at every zoom. |
| Per tile | The same, without the pixel limits. |
| Uniform | One radius, in metres, for every tile. |
| Screen pixels | Every point the same size on screen, wherever it is. The most exact and the heaviest: all point tiles are re-evaluated whenever the view moves. |
If points look too sparse or too fat, change Point scale first.
Colours
Points keep their own colour. To colour them by class or by intensity, see
Nodes: Metadata; to drape a map on them, see
Raster overlays. The OGC3D Points node group is shown in
Nodes: Geometry.
How it works
- A point tile is a mesh with vertices only, a
Colcolour attribute, and the sharedOGC3D PointsGeometry Nodes modifier. - Attenuated uses 0.6 × the tile's mean point spacing, measured at the distance from the eye to the view's orbit centre (from the camera to the tiles when rendering). One distance for the whole scene gives every tile of a level the same radius, as CesiumJS' point cloud attenuation does. The sizes are recomputed only when that distance changes by more than 15 %, so tiles are re-evaluated while zooming, not while orbiting.
- Screen pixels reads the eye position from the
OGC3D ViewEmpty, per point. - Disc splats keeps the glTF normal as the
ogc3d_normalattribute.