Cutting on action
Also known as: action cut, cut on movement
Placing the cut in the middle of a movement rather than between movements, so the motion carries across the join and hides it.
Updated
What it means
A plain Cut during stillness is visible. A cut during movement is not, because the eye is following the motion rather than examining the frame. Attention is spent on direction, not on the join.
The movement has to continue on the other side. A rider leaving frame in one shot and arriving in the next works; the same rider stationary in the second shot does not.
It is the mechanism behind most invisible editing, and the reason a well-cut sequence feels like one continuous thing rather than a set of clips arranged end to end.
How it works in practice
Two angles of one movement is the ideal case. A takeoff from a wide and a landing from a Chest mount, cut mid-air, reads as a single action from two positions.
The frames have to match. Position in frame, direction of travel and speed all need to continue, which is why the 30-degree rule and this technique are usually applied together.
A Match cut is the stylised version. The movement continues but the location changes, which is the same principle used deliberately rather than invisibly. Both rely on a viewer completing the movement.
What to watch out for
Mismatched speed is the usual failure. Two angles of one jump recorded at different frame rates or stabilised differently do not continue into each other, and the cut draws attention to itself exactly where it should disappear.
Cutting too early or late loses the effect. A few frames in either direction turns an invisible join into a stutter, which is why these cuts are trimmed rather than placed.
Automatic assembly cuts on content rather than on movement. Boundaries land where Scene detection found a change, so a sequence assembled that way needs its joins checked by eye before anybody calls it finished.
How RawClip handles it
Cut points are chosen from the action inside each clip rather than at arbitrary intervals, so joins tend to land where something is moving. Coverage of one moment from two angles is what makes that work, which is an argument for a second camera.