Where the head is turned, relative to whatever the pose aims at.
Three numbers say where the camera is; these two say where it looks from
there. They are separate because the orbit arm aims at one point by
construction — lookAlong(−offset, up) is the target's center, always — and
a planetarium in which the only thing you can look at is the middle of the
subject cannot show you a moon beside a disk, a limb, or the sky.
An offset rather than a replacement, so the composed aim stays the thing it
is measured against: a shot solves angles and a distance, and the viewer
turns their head from there. Zero is the composed aim itself, and
observerPose returns the identical pose for it — the compositions are
fitted against that number and a rewrite that quietly went through the
quaternion product would move every one of them by an ulp.
Members
pitchreadonly
Above the camera's own right axis. Clamped by ELEVATION_LIMIT.
yawreadonly
Around the camera's own up axis. Unbounded — it wraps by construction.