FUSION · NODE REFERENCE
dTransform
Deep Transform · screen-space position and sample-Z adjustment
Deep Transform [dXF] repositions and scales a Deep image in the image plane while independently scaling and shifting its sample-Z depth before flattening.
dTransform
Overview
What dTransform does
Purpose. dTransform works on Deep Image streams, preserving multiple per-pixel depth samples while allowing spatial positioning, image-plane scaling and independent Z adjustments. It is not simply another 2D Transform node.
Edition. The Deep Image compositing toolset arrived in Resolve 20 Studio. This page documents the tool; it does not imply full multisample EXR processing in the free version.
Source screenshot. The genuine archived dtransform-controls.png (397×362) shows Scale Z=1.0, Translate Z=0.0, Sample, Center X/Y=0.5, Pivot X/Y=0.5, checked Use Size and Aspect, Size=1.0 and Aspect=1.0. These are observed UI values, not a recommended preset.
Flow contract. The orange Image input expects Deep samples; the optional blue Effect Mask restricts the operation. Output remains Deep. Insert DeepToImage before feeding a normal 2D Merge or MediaOut.
Two spaces. Center/Pivot/Size adjust the onscreen projection in X/Y. Scale Z and Translate Z change sample-depth values. Increasing Size does not scale sample-Z depth, and Scale Z need not increase apparent image size.
Scale Z. The reference manual describes expansion or compression of the depth range. 1.0 is the neutral captured setting. Check units and camera conventions first when merging Deep elements from different renders.
Translate Z and Sample. The captured Z offset is 0.0. Sample is a Viewer picker for a Z value, not an interpolation algorithm or a separate sample mode. Never assume a universal positive/negative near/far convention without testing your source.
Center. Center X/Y moves the image-plane projection in normalized frame coordinates. If you need simultaneous X/Y and Z movement, isolate and check each operation independently before combining them.
Pivot. A scaling anchor influences the onscreen expansion point, but does not reconstruct the original 3D camera or rescale scene geometry behind the rendered Deep samples.
Two scaling modes. The captured checked Use Size and Aspect exposes Size and Aspect. The manual also documents separate Size X and Size Y values in the alternate, unchecked mode. That alternate UI is NOT shown on the archived screenshot.
Versus dResize. dResize changes the Deep canvas Width and Height; dTransform changes image placement and sample depth. Neither automatically turns Deep samples into normal RGBA.
Linear compositing. Blackmagic states that Deep compositing needs a linear processing space. Confirm matching source color spaces and alpha handling before blaming edge artifacts on depth translation.
Optional mask. Effect Mask limits the region affected by the transform, not a second Deep layer or a camera socket. Inspect feather and Deep overlap near the mask boundary.
Finishing. A typical chain is Deep source → dTransform → dMerge → DeepToImage → ordinary 2D Merge over video → MediaOut. DeepToPoints is a diagnostic scene output, not a 2D image replacement.
Actual multisample evidence. ImageToDeep is useful for teaching contracts but does not stand in for a complex overlapping multisample Deep EXR; visual occlusion fidelity requires genuine test assets.
Verification limits. Graph/type tests can establish consistent examples and article content, but not GPU rendering, native .comp import, EXR resampling, depth sort correctness or target-build options. These remain pending in Resolve Studio.
When to use it
Offset Deep sample depth before dMerge
Match Z scale across Deep render sources
Reframe a Deep card without early flattening
Scale Deep imagery on screen while preserving depth samples
Place Pivot at a desired image-plane anchor
Stretch axes independently via alternate Size X/Y controls
Constrain the operation using a soft Effect Mask
Prepare a Deep branch for dHoldout or dRecolor
Evaluate sample occlusion after DeepToImage
Maintain a linear Deep composite before final 2D treatment
dTransform in context
Screenshots referenced by this article. Open an image to inspect it at full size.
Example for dTransform.
Original image