Locator3D

Locator3D's primary job is not merely to create an empty helper in the 3D Viewer. Blackmagic describes it as a way to transform a point in 3D space into 2D coordinates that can drive other controls through expressions or modifiers. A common production use is attaching a 2D callout, mask, label, or graphic to a moving point in a 3D scene.

Locator3D has two external Flow inputs and both use classic Fusion scene3d data. The orange SceneInput is required. The green Target input is optional. Treating Locator3D as a generator with no inputs hides the projection context the tool actually needs.

SceneInput must contain a camera used for projection. A reliable setup places Locator3D after the Merge3D that already combines the camera with the relevant 3D scene. A Viewer viewpoint by itself does not provide the same documented projection context.

Camera in the Controls selects the specific camera from the incoming scene that defines screen space for calculated Position. In a scene containing multiple cameras this choice is part of Locator's math, not merely a Viewer preference.

Width, Height, and Pixel Aspect also participate in the 3D-to-2D mapping. They should match the image produced by the corresponding Renderer3D. A correct 3D point paired with the wrong frame geometry can still produce a systematic 2D offset or scale error.

The captured current Resolve UI labels the automatic frame-size option as Auto Resolution. The reference manual describes related behavior as Use Frame Format Settings. This article treats that as a version/UI-label boundary: the role is documented, while the exact label, default state, and target-build behavior remain a live-verification item.

Calculated Position is not an RGBA image and not a normal Flow-cable output. It is a numeric/point control source. For example, right-click a 2D mask Center, choose Connect To, then select Locator3D > Position. That control-level link is how 3D motion drives a 2D parameter.

The optional green Target input accepts scene3d. With Target connected, Locator is positioned from the target scene/object transform center. The reference behavior also makes Locator Translation XYZ offsets operate in the target's local coordinate space instead of acting like an unrelated absolute world position.

Target input and Transform > Use Target are separate mechanisms. The green Target input supplies a positional base from another 3D object/scene. Common Transform > Use Target is a look-at orientation feature. Do not merge the two concepts simply because both use the word Target.

Sub ID can address a supported subelement, with documented examples such as an individual Text3D character or a Duplicate3D copy. Exact indexing for complex imported assets should still be confirmed on the target Resolve build.

Size and Color primarily control Locator's onscreen crosshair. Size is not world-space Scale and does not change the projection math. Color is useful for distinguishing several locators in a Viewer but does not turn the marker into a normal material workflow.

Make Renderable lets the Locator crosshair participate in the documented OpenGL/Hardware rendering path. The Software Renderer cannot render these line primitives, so it is not a valid positive baseline for a renderable Locator marker.

When Make Renderable is active, conditional Unseen by Camera can keep the Locator useful in 3D viewers while excluding the marker from the camera render. That display/render rule should not be used as a switch for calculated Position.

Matte controls belong to common 3D matte/holdout behavior. Is Matte is not ordinary 2D alpha = 0; it participates in scene-depth holdout semantics. Conditional Opaque Alpha and Infinite Z belong to the matte state even when they are hidden in a screenshot captured with Is Matte disabled.

Normals/Tangents Scale and Show controls are common 3D diagnostic overlays. They do not participate in screen-space projection and do not recompute surface vectors. Use ReplaceNormals3D when the task is actually to rebuild normals or tangents.

The Transform tab controls Translation, Rotation, Pivot, and Scale of the Locator object itself. For basic 3D-to-2D tracking, Translation or the Target input is usually enough; Rotation/Pivot/Scale matter when the Locator also participates in a larger 3D rig or debug representation.

Locator3D's normal Flow Output remains classic Fusion scene3d for Merge3D, Renderer3D, or other 3D stages. This is independent of calculated Position. Directly feeding Locator3D.Output to MediaOut is a type mismatch because Renderer3D is the scene3d-to-image boundary.