ImageToDeep

Start by identifying the source. A flat image with alpha and a genuine multi-sample Deep EXR contain different information. ImageToDeep lets the former join a Deep composite, but it does not reconstruct every surface behind the foreground. A smoke plate may become a depth-positioned card rather than a true Deep volume.

Separate three decisions: how Z is established, how it fits the scene’s convention, and where the image belongs in the frame. Specify Z selects the assignment path; Z Scale and Z Offset help match depth placement. Center X/Y controls two-dimensional registration. Keeping those decisions separate makes a wrong result easier to diagnose.

Depth correctness is usually judged against other surfaces. Combine the prepared element with a compatible Deep scene in dMerge and place the ordinary image preview after DeepToImage. An isolated layer can look similar at several depths even though its occlusion inside the scene changes.

Deep compositing requires a linear working space, aligned frames and a known depth convention. ImageToDeep does not automatically repair camera, timing or color-management mismatches. The manual also permits automatic 2D conversion in other Deep tools; using an explicit converter provides a visible place to control and inspect the assignment.

The source PNG shows five Controls entries. The additional guide entries are data and compositing checks. Verify the actual Z-assignment interaction, the result of Sample and rendered frames in your installed Resolve build; the examples on this page are not exported .comp files.