The gizmo & direct editing

Look at an entity, pinch to select it, and it's yours — grab it directly, or summon a transform gizmo for precision. Everything you'd do with a mouse on the Mac, you do here with your eyes and hands.

Selecting in space

Every entity in your scene is gaze-pickable: there's no edit mode to switch on; the workspace is always an editing surface. Look at an entity (it highlights) and pinch to select it. The selection routes everywhere at once: the inspector shows the entity's transform, and the timeline scrolls to its clips.

What happens in space is up to you. By default your hands own the selection directly: grab and move it, no gizmo in the way. When one axis matters, the mode picker in the transport bar (Move / Rotate / Scale) summons that handle family around the selected entity. One family shows at a time; tap the active mode again to put the gizmo away and go back to direct manipulation.

The gizmo sizes itself to encompass whatever you select, so it hugs a small prop and wraps a large one. It's also capped at roughly 0.3 meters across, so selecting a room-sized backdrop won't summon a gizmo that dwarfs you. The handles stay a comfortable arm's-length tool no matter the subject.

Reading the handles

Whichever mode is up, a grab bar rides beneath the handles: pinch-drag it to free-move the entity without leaving the gizmo. You never have to drop back to direct manipulation just to reposition.

Each drag is frozen-frame stable: the moment you pinch a handle, the gizmo locks its axis frame for the duration of the gesture. Walking around the entity mid-drag, or the entity animating underneath you, never bends the axis you're pulling along. The drag tracks exactly the line or ring you grabbed.

Direct drags: the default

With no gizmo mode active, manipulation is direct, and it's always on. A one-handed pinch-drag moves the entity through space, with wrist rotation turning it as you go. Bring in your second hand and pinch with both to scale. These direct manipulations write the same transform edits as the gizmo: the gizmo is for precision, direct drags are for speed.

Auto-key: drags that write animation

The gizmo is animation-aware. With the record button (auto-key) enabled in the transport bar, a gizmo drag on an already-animated entity writes keyframes for exactly the channels you changed, at the current playhead time. Slide an animated entity along x, and only its x-position channel gets a new key; its rotation and scale curves are untouched.

Auto-key deliberately never keys a channel that has no animation yet. Nudging a static prop with record on repositions it; it doesn't sprout an accidental one-key animation track. To start animating an entity, set an explicit pose key first (see keyframes & auto-key).

Motion trails: grab the path itself

Animated entities draw their motion as a trail through space: the actual path the entity will travel during the segment. Along the trail sit grabbable orbs, one per position keyframe. Pinch an orb and drag it to a new spot, and the keyframe underneath it moves with it. The curve reshapes live, and the trail redraws as you drag.

This is the most spatial way to author movement in Chapter Vision: instead of scrubbing to a time and adjusting numbers, you sculpt the flight path with your hands, standing where the audience will stand.

Pose keyframing from where you stand

The transport bar's keyframing controls capture poses in place. Position an entity (gizmo, direct drag, or inspector numbers), scrub to the moment you want, and set a key: the entity's full pose (position, rotation, scale, opacity) is recorded at that time. Move, scrub, key again, and the entity animates between your poses. Because you author from inside the scene, you're always judging the move from a real viewpoint, not a proxy viewport.