Purpose

Depth Composite combines two rendered layers using their depth images. The background establishes every pixel. Foreground replaces it only where foreground depth is strictly nearer, with premultiplied-alpha blending for partial foreground coverage.

Use

Place Depth Composite in the Geometry view, under Render. Render two object collections from the same perspective camera at the same resolution. Use the same source frame for both.

Connect the first Path Trace’s beauty and depth to background and background_depth. Connect the second renderer’s beauty and depth to foreground and foreground_depth. Start with opaque objects that overlap in the camera view so you can check which layer wins.

Inputs and outputs

All four inputs are required. Color inputs are float32 RGBA; depth inputs are float32 single-channel images. The outputs are color and depth, using the common input dimensions.

Keep each color image paired with its own depth. The inputs must also carry matching Geometry camera and depth information. An arbitrary depth image is not interchangeable with Path Trace’s world-unit ray distance.

Controls

There are no operator-specific parameters. Input order and depth determine the result. Adjust camera, resolution and scene selection on the upstream renderers.

Equal depths keep the background, including equal sky misses. A strictly nearer foreground with alpha between 0 and 1 blends over the background as premultiplied color. The output depth then takes the foreground depth.

Notes

Depth Composite checks dimensions, camera, projection, depth convention and source frame. It reports detected mismatches and does not reproject images from different cameras.

A foreground with alpha 0 is not automatically skipped by the depth test. If its depth is nearer, it still replaces the pixel. Empty foreground pixels need the appropriate miss depth; alpha alone is insufficient.

Composite in linear color before display conversion. The operation keeps one depth per pixel and is not a deep-image compositor.

Ports

Inputs

• background image required
  RGBA
• background_depth image required
  single channel
• foreground image required
  RGBA
• foreground_depth image required
  single channel

Outputs

• color image required
  RGBA
• depth image required
  single channel