Nodes

Object Solve

Solve a moving object against the camera solve.

Works out how something moves relative to a solved camera, so a car, a prop or a person can carry CG.

Inputs

You need Scene plus at least one of the other two. Give it both and they merge: the automatic solve provides bulk, the hand tracks provide points you can identify on your model.

Output is Object Data.

Getting The Object Solve

Mask the object with AI Mask, then run two Auto Track nodes off the same mask: one with Mask Mode set to Exclude for the camera, one set to Limit To for the object. No inverted mask needed.

Settings

Object Name names the solve. It also has to match the target name in the 2D/3D Constraints panel if you want to align a mesh later.

Min Track Length ignores object tracks shorter than this many frames.

Robust Loss (px) is the outlier threshold.

Sparse-Frame Damping applies a constant-velocity assumption to frames with very few tracks, which steadies moments where the object is barely visible.

Object Scale

Solving an object separately leaves a real ambiguity: a small object close to camera looks identical to a large one further away. The footage cannot tell you which.

Object Scale resolves it. Dragging scales the object and moves it toward or away from the camera together, so the screen match never changes while the depth does. Drag until it sits believably against your point cloud. It updates live once the object is in the scene.

Placement

Lineup Object picks a mesh to parent under the solved empty automatically at import. Often enough on its own when the origin already matches.

Lineup Points also creates the solved object points as a vertex cloud.

For precise placement, link hand tracks to mesh vertices and use 2D/3D Constraints. Those tracks must be wired into Tracks here before you solve.

See Also

Object Tracking.