Beyond Z-Depth: Unlock Smarter Compositing with Deep Images in Resolve

Dive into OpenEXR 2.0 for Seamless Layering of Transparent Elements—Without the Usual Headaches

Have you ever struggled with jagged edges or inaccurate depth when compositing semi-transparent elements?

In this Insight, we’ll introduce you to deep image compositing using OpenEXR 2.0 files, now supported in DaVinci Resolve and Fusion since version 20. You’ll discover how these multi-sampled images outperform traditional Z-depth channels, making your workflows more precise and efficient—especially if you’re dipping into high-end VFX pipelines.

Whether you’re a colorist curious about emerging tech or a compositor handling complex 3D renders, understanding deep images equips you to join conversations with VFX teams. We’ll break down the basics, highlight key tools, and show practical examples so you can experiment right away.

“Deep image compositing allows for more accurate merging of transparent elements by storing multiple depth samples per pixel. Switching to a deep image workflow, we can bring flat elements into the deep image workflow…and use a deep merge tool to composite with correct depth layering.”

Bernd Klimm, Compositor

Why Deep Images Change the Game for Compositors

As we explore further, you’ll see why traditional depth channels fail in real-world scenarios.

Starting with a simple chessboard render featuring depth of field, motion blur, and transparency, the Insight reveals how a flat Z-channel struggles to handle overlapping pixels, leading to blocky, inaccurate results that no amount of smoothing can fully fix.

Switching to deep images, we can store multiple depth samples per pixel, preserving transparency across layers. This means cleaner merges, even with complex elements like fog or semi-opaque overlays. In Fusion, you can toggle between flat and deep modes in the Loader or MediaIn nodes, giving you flexibility to integrate these files into your existing setups.

Building on that, this Insight demonstrates how to convert flat elements into deep ones using dedicated tools. If you’re working with 3D renders from engines like Arnold or Renderman (note: free options like Blender’s Cycles don’t support this yet), deep images streamline your pipeline by embedding all the spatial data needed for flawless compositing.

Setting Up Your Deep Image Workflow

Ready to try it? We’ll walk through the essentials, starting with identifying deep files in Fusion—they show as “Deep 32” in the viewer. For a quick test, load an OpenEXR 2.0 sequence and switch to flat mode to access the Z-channel via a Copy AUX node.