Transform3D

Transform3D does not generate geometry and does not process a 2D image. Its one required orange SceneInput accepts classic Fusion scene3d from Shape3D, Text3D, ImagePlane3D, Merge3D, imported geometry, or another compatible 3D tool. The Scene output remains scene3d.

Its main production role is to separate spatial animation from the node that creates the object. Instead of mixing geometry construction and motion inside Shape3D or Text3D, keep the source stable and move the completed branch with a dedicated Transform3D layer.

Transform3D applies an additional matrix on top of upstream transforms. Two Transform3D nodes in series are therefore not redundant: the first can own a local rotation while the second moves the already-transformed branch as a parent-like level.

Node order matters. Shape3D → Transform3D_A → Transform3D_B and Shape3D → Transform3D_B → Transform3D_A can produce different results because Translation, Rotation, Pivot and Scale do not generally commute. Name transform layers by responsibility in complex rigs.

Translation X/Y/Z moves the entire incoming scene branch in 3D space. It is a spatial offset rather than a 2D Center value or a camera control. When Transform3D sits after Merge3D, Translation moves the complete merged subtree.

Rotation Order chooses how Euler rotations are evaluated: XYZ, XZY, YXZ, YZX, ZXY or ZYX. With only one active axis the difference can be invisible, but combined X/Y/Z rotations make order part of the final orientation and animation.

Pivot X/Y/Z defines the point around which Rotation and Scale affect the incoming scene. Pivot is not an additional Translation control. An offset pivot combined with Rotation creates orbit-like motion rather than a simple spin around the original center.

Scale is linked by Lock X/Y/Z by default. While locked, one Scale value changes all axes uniformly. Unlocking exposes separate X/Y/Z Scale controls for non-uniform transforms. This scales the whole branch rather than changing primitive resolution or subdivision.

Use Target enables target-based orientation. Blackmagic documents XYZ Target Position controls after enabling it; the object or branch is oriented toward that point while Rotation becomes a relative offset from the computed target orientation. Translation still defines the source-branch position.

Target Position is an aiming coordinate rather than a second SceneInput or a separate Flow object. In a production rig those coordinates can be linked to another 3D element so a card, title or other oriented branch keeps facing a moving target.

Invert Transform is visible in the bundled Resolve 21 Inspector capture. Blackmagic's Fusion tool reference describes it as inversion of the position, rotation and scaling transform. It is useful in compensation rigs where a transform and its inverse cancel at the correct hierarchy level.

Do not treat Invert Transform as simply multiplying X/Y/Z values by -1. A matrix inverse accounts for the composed translation, rotation, pivot and scale; with multiple axes or an offset pivot, manual sign changes are not equivalent.

Import Transform imports transform data rather than geometry. The current Blackmagic manual lists LightWave Scene (.lws), Max Scene (.ase), Maya ASCII (.ma) and dotXSI (.xsi). Use the appropriate 3D import workflow when you need geometry, lights, cameras or the complete scene itself.

The Translation, Rotation and Pivot Pick controls provide interactive Viewer routes to the same spatial tasks. Blackmagic also documents onscreen transform modes and Q/W/E shortcuts for translation, rotation and scale.

Viewer gizmos and Inspector values should describe the same transform layer. While Lock X/Y/Z is enabled, do not treat a single-axis scale gesture as an independent non-uniform scale workflow; unlock the axes first and verify the separate controls.

Transform3D is useful in UV/texturing pipelines: geometry can be processed in neutral coordinates by UV Map 3D or other geometry tools, then placed in the final scene with a downstream Transform3D. This separates surface preparation from shot placement.

It is also useful after Merge3D when several objects have already been assembled into a group. A single downstream Transform3D becomes a group-level parent-like motion layer without duplicating identical keyframes across every child object.

Do not confuse Transform3D with transforms available inside other 3D tools. A dedicated Transform3D creates an explicit, named responsibility that is easier to reorder, duplicate, bypass or invert in a technical rig.

Do not confuse Transform3D with the 2D Transform node. After Transform3D the data is still scene3d: there are no RGBA pixels, image canvas dimensions, 2D edge behavior or an Effect Mask. Normal 2D Merge/Glow/MediaOut become valid only after Renderer3D.

The Transform3D Scene output can feed Merge3D, Renderer3D, another Transform3D, UV/geometry processors and other classic Fusion 3D nodes that accept scene3d. The cable type does not change when the matrix becomes more complex.

For troubleshooting, isolate layers in order: topology/data type, Translation, Rotation Order and angles, Pivot, Scale, then Target or Invert. Changing every group at once easily hides a hierarchy mistake behind a coordinate adjustment.

A useful diagnostic baseline is a neutral transform: Translation 0/0/0, Rotation 0/0/0, Pivot 0/0/0, uniform Scale 1, Use Target off and Invert Transform off. Change one group at a time, then add the next hierarchy layer only after the previous one is understood.

The bundled Resolve 21 screenshot confirms the current Transform3D Controls layout with Translation, Rotation, Pivot, Scale, Use Target, Invert Transform and Import Transform. Conditional Target Position and unlocked X/Y/Z Scale are documented by Blackmagic and remain items for live target-build evidence when final certification is performed.