Spherical Domain

The Spherical Domain captures the scene into a native cubemap target and derives the supported spherical representations from that capture.

Capture

The cubemap contains the six directions required to represent the spherical scene. sceneRender() can therefore be invoked multiple times during a single Processing frame while the cubemap is captured.

This is why mutable state must advance in update() when all faces must observe one coherent state.

Projection siblings

Domemaster, Equirectangular and Skybox are sibling final projections derived from the cubemap. They should not be documented as a serial conversion chain when the implementation can project them directly from the shared cubemap. Skybox retains the qualified cross slots but maps each face with true equi-angular tangent coordinates (EAC).

The six faces share one monotonic nanosecond capture timestamp. A publication barrier remains closed while the reusable OpenGL target renders them sequentially and opens only after all faces and mipmaps complete. This is an atomic logical batch, not a sleep or an assertion that one OpenGL context can rasterize six cameras literally at once.

Pitch/Yaw/Roll belongs to the shared spherical orientation. Domemaster additionally applies its FOV and Size% calibration controls.

Each cubemap face starts transparent, and its alpha is sampled unchanged by Equirectangular and Skybox. Domemaster multiplies sampled alpha only by its derivative-based circle coverage: pixels outside the circle remain transparent, a smaller Size% leaves the newly exposed area transparent, and Size%=0 yields a fully transparent final target. Missing/failed projection inputs clear the destination rather than retaining an earlier frame.