The camera is no longer bolted to the player: it is set in the world, on a rail you trace and a timeline you cut.
The tool
Viewpoints are set right where you stand. The mod draws a continuous curve from them, visible in the world, editable point by point. Adding an anchor in the middle of a move does not distort it: the motion stays what it was, with one more handle to correct it.
Speed is read from the distance actually travelled, not from anchor numbering, which removes the jolts wherever two anchors sit close together. And durations are counted in seconds, never in frames: a shot lasts the same time on any machine, whatever frame rate it manages.
Nothing is faked on screen. Before the take, the state of the world is captured; on playback it is put back as it was, blocks and attached data included, then the gesture is performed again for real by a stand-in. What fell falls again, what was placed is placed again, on the same second.
So an action is performed once, but filmed as many times as needed, from a different angle on every take. That is what makes it possible to cut a scene the way film does, without ever asking anyone to repeat the exact same move.
An actor carries a name, a look, and can take the shape of one of the game's creatures. It doubles the character rather than replacing it: play carries on while the stand-in performs its take.
Several actors hold the same shot, each replaying what was recorded for them. A scene is therefore no longer limited to what one person can do in one go.
It all fits in one item, the trunk, and in a stage mode: while it is on, the game stops playing under your fingers and every key goes to the tool in hand. The camera, the sight, the boom, the lamp, the floor mark and the desk share the work between them.
Gestures follow the same grammar from one tool to the next, and what is done in the world can be undone. Nothing goes through chat: what gets tuned gets tuned where it is seen.
Takes are sorted into shots, shots are strung into a film. A panel holds the list, the curve is sampled at the step you choose, and the whole world can run in slow motion to catch what moves too fast.
Locked-off cameras, a rail that follows an actor and ready-made paths cover the common cases without starting from a blank page.