PointCloud3D

The most important PointCloud3D workflow is 3D camera tracking. A CameraTracker solve can export a Camera3D, PointCloud3D, GroundPlane, Merge3D and Renderer3D setup. The points typically represent reconstructed feature tracks and help establish the ground plane, scale, and placement of CG relative to live action.

PointCloud3D does not perform tracking or a solve by itself. CameraTracker analyzes the 2D plate and reconstructs the camera/points; PointCloud3D stores and displays the resulting 3D positions. Connecting a MediaIn directly to PointCloud3D therefore does not convert footage into a point cloud.

The node has one external SceneInput. It is an orange classic Fusion 3D scene input, not an image input. A wire is often unnecessary in CameraTracker/export and Import Point Cloud workflows because the point data can already live inside the node. The output remains classic Fusion 3D data and normally feeds Merge3D.

Controls mainly determine how the points appear in the Viewer. Style switches between Cross and Point. Density controls the probability of displaying each point. Default Color supplies a uniform marker color, while Use Per-Point Colors uses color data stored per point when the source cloud actually contains it.

Density has a precise meaning in the Blackmagic reference: 1 displays every point and 0.2 displays about every fifth point. It is not a point generator, subdivision amount, or generic intensity control; it only thins the displayed set.

In Cross mode the current Resolve 21 Inspector shows Lock X/Y/Z and Size. With Lock disabled, the X/Y/Z crosshair arms can be sized independently. The Resolve 21 Point-style screenshot stored with this project instead exposes Size and Antialiasing for smoother point markers.

Import Point Cloud reads external scene data. The current reference lists Maya .ma, 3DS Max ASCII .ase, LightWave .lws, and Softimage XSI .xsi. This is 3D/point-data import rather than raster image loading.

Make Renderable separates reference-only use from deliberately drawing the point markers into a final render. With it off, the cloud is still useful as a 3D guide. With it enabled, markers can be included by the OpenGL Renderer3D path; Blackmagic explicitly notes that the software renderer does not support rendering the visible crosshairs for this node.

When Make Renderable is enabled, Unseen by Camera becomes available. It keeps the cloud visible in the viewers while excluding it from the camera render, which is useful when the artist needs reference markers but the audience must not see them.

Transform and Settings are common classic-Fusion-3D tabs. Translation, Rotation, Pivot and Scale move the whole point-cloud object; they do not rewrite each internal point position. Controls/Size and Transform/Scale are different: Size changes marker size while Scale changes the entire 3D object.

PointCloud3D also has a useful Viewer contextual menu. Find locates points by name, Rename renames selected points, Delete removes them, and Publish exposes selected point coordinates as Inspector controls. Selected points can also create a Shape, ImagePlane or Locator, which turns the cloud from a passive reference into a practical placement aid.

The wire remains scene3d until Renderer3D. Merge3D assembles PointCloud3D with the camera, ground plane and other geometry. Renderer3D is the boundary that turns the scene into a normal image for Merge, color correction, glow, grain and MediaOut.