Grayscale map encoding pixel distance from camera — white near, black far. Essential for post-production depth effects and 3D scene recovery.
In digital compositing, in addition to the RGB image, you need information about spatial depth – this is precisely what a Z-depth image (or depth map) provides. Each pixel is assigned a grayscale value that encodes its distance from the camera: white means close, black means far away. This is not an optical phenomenon but pure geometric information – essential for simulating focus effects afterward or positioning 3D layers correctly.
On set or in 3D software, you generate these maps parallel to the beauty plate. In production, they usually come from the rendering engine (Arnold, RenderMan, V-Ray) or from a depth pass calculation for live-action footage (photogrammetry, LiDAR scan, or manual rotoscoping). The technical implementation varies: Linear Depth stores the actual distance in world units, Normalized Depth compresses the range to 0–1 for better bit-depth efficiency. Some pipelines also use logarithmic encoding to capture near and far ranges uniformly.
In post-production, the Z-depth image becomes the control map for Depth-of-Field effects (see Focal Plane Compositing). You define a focal plane – all pixels with Z-value X remain sharp, everything before and after it is blurred with variable blur. Modern compositing software (Nuke, After Effects with plugins) can calculate this photorealistically: the amount of blur scales continuously with depth. This also allows you to shift a focal plane afterward – focus racking in the edit, without reshooting.
A common beginner's trap: linear Z-maps can lead to quantization artifacts in large depth ranges. If your subject extends from 1 meter to 500 meters and you only have 8-bit accuracy, banding errors will occur in the blur transitions. Therefore, it's better to export in 16 or 32 bits and use normalization techniques that intelligently distribute the available value range. Also important: the Z-map must be perfectly registered with the RGB image – any sub-pixel shift will lead to halo artifacts around edges. For motion footage, you need a separate depth map for each frame, not just a static one.