Coverage Node

Coverage isolates the coverage channel and lets you handle it independently from the other data. Its output can be used as an alternative matte, to drive other operations, or for visual debugging.

Coverage is typically more precise than the standard alpha because it includes multi-layer fragments: when several transparent surfaces overlap, the summed alpha can differ significantly from the real geometric coverage.

• Effects Library: Tools > Position > Coverage
• Quick add: right-click in the Node Editor → Add Tool > Position > Coverage

Alpha

Per-pixel opacity after all the compositing done inside the 3D engine. Includes the effect of every overlapping transparent surface.

Coverage

The geometric percentage of the pixel actually covered by the surface, regardless of internal transparency. Precise for antialiased edges.

The practical difference

On an antialiased edge of a 100% opaque object: alpha = coverage. On a pixel containing a glass edge + a background behind it: alpha reads the final attenuation, coverage still says 100% (the pixel is fully covered by something).

• Input (orange) - a multichannel image containing a Coverage channel (typically a multichannel 3D EXR).
• Effect Mask (blue) - optional.

Premium antialiasing in compositing:
Multichannel EXR → Coverage (extract)
                  → Channel Booleans (replace alpha with coverage)
                  → Merge with the background → MediaOut

Visual debug of coverage vs alpha:
EXR → Coverage (visualize)
   → MediaOut (next to the alpha for comparison)

Simplified multi-pass deep compositing:
Pass1 (with coverage) → Coverage
                     → Merge with Pass2 driven by coverage
                     → MediaOut

Channel (default: Coverage)

The source channel to analyze. Coverage by default, but it can point to other channels depending on the naming of the source EXR.

Output as Alpha (default: On)

When enabled, places the coverage in the output alpha channel for direct use as a matte.

Threshold (default: 0)

The threshold below which coverage is treated as zero. Lets you exclude pixels with very low coverage (stray traces).

• Optimal multi-render antialiasing - combine several 3D renders with their coverage = edges without stray aliasing in compositing.
• Simplified deep compositing - for artists without access to full deep images, coverage offers a useful in-between.
• Debugging edge artifacts - viewing the coverage alone helps spot poorly covered pixels or holes.
• Precise matte for a transparent stack - glass + reflection + object behind = coverage gives the total coverage, alpha only the attenuation. Choose based on what you need.
• Correct premultiplication - for workflows premultiplied by coverage rather than alpha.

• Check the source channel - the exact name varies by 3D engine. Fusion detects "Coverage" by default, but some exports may call it "coverage" (lowercase) or "object_coverage".
• Coverage ≠ Alpha on transparent objects - for an opaque object they are identical; on overlapping transparent objects, they can differ significantly.
• Threshold for cleanup - removing coverage < 0.01 = nearly transparent pixels that are often stray.
• Multichannel EXR required - coverage does not exist in simple formats (PNG, JPG). Multipart or multilayer EXR is mandatory.
• Do not confuse it with Z-depth - coverage is a 2D measure of coverage, not a 3D distance. For depth, see Z to World or Volume Fog.
• For full deep workflows, use dedicated tools (Nuke DeepRecolor, Houdini Deep Compositing). Fusion's Coverage covers the standard cases.