The title is drawn on the canvas; axis names become the slider labels for the points below.
Fields depend on the mode: orbits takes a ring and a position; axes takes values on your axes; well takes a position, a weight (minus is a hill) and a funnel radius.
items: label, x/y/z (−1…1), weight ±, radius, ring; links: label pairs. One dataset serves every mode.
A keyframe stores the whole state: mode, camera, parameters, data. Dur is the seconds to the next keyframe; ▶ eases from the current view to that keyframe.
Only the canvas is recorded, the panel stays out of the video. The file downloads when the pass ends.
Config is parameters plus data. The full scene state is reproducible.