depth32float-stencil8

Request depth32float-stencil8 in the device’s requiredFeatures before creating textures with this format. It gives a depth-stencil attachment an explicitly defined 32-bit floating-point depth component alongside an 8-bit stencil component. Depth testing resolves visibility while stencil values can mask portals, outlines, or selected regions of a render.

The feature is useful for engines that need a predictable floating-point depth representation while retaining stencil operations, including large scenes and reversed-depth renderers. The extra precision can cost memory and bandwidth; do not assume the physical allocation is simply the sum of the advertised component bits.

If it is unavailable, depth24plus-stencil8 provides a core combined attachment with an implementation-chosen depth representation. If stencil is unnecessary, core depth32float keeps the explicit floating-point depth format. Depth and stencil aspects also have different sampling and copy rules, so use the appropriate aspect when making views or transferring data.