Planimetric staging is a hold, not a snap
The flat, dead-on, perfectly symmetrical shot is instantly recognizable and it looks like it should be easy: put the camera on an axis. Doing only that produces something that reads as stiff rather than as composed, and the missing piece is not in the geometry.
The left camera is always moving and always between angles, so no arrangement of the room is ever the arrangement. The right one arrives at a compass point and stops, and while it is stopped the layout is a picture with a front, a back and a symmetry.
The three rules, and only one of them is about angles
- The camera axis is perpendicular to a wall. Ninety degree increments, no three-quarter view. This is the part everyone gets from looking at a reference frame.
- Movement is along an axis or it is a cut. Dolly straight in, track straight across, or cut ninety degrees. Never arc, because an arc passes through every angle the discipline exists to avoid.
- Hold on the composition. The style is a series of arrangements the viewer is given time to read. A perfectly staged frame held for six frames is not staged at all.
That third rule is the one that gets dropped, because it costs screen time and the other two cost nothing. It is also the one doing the work: the appeal of the look is that somebody arranged something and then let you look at it.
Why it is worth so much in a field renderer specifically
Two reasons, and both are practical rather than aesthetic:
- Axis-aligned shapes are cheaper and cleaner. A box on an axis has an exact, trivially antialiased edge with no rotation in the field. A room full of arbitrary angles is a room full of rotations evaluated per pixel.
- A held frame can afford much more per pixel. If the camera stops, the previous frame is still valid, so accumulation over time becomes free. Twenty samples a frame for twenty frames is four hundred samples, and the style is the reason it works.
So the constraint pays for itself twice, which is unusual. Most style decisions cost performance.
Compass-point editing
The cutting rule that goes with the staging: every cut is a ninety or a hundred and eighty degree turn, never a small adjustment. A twenty degree change of angle is the one thing to avoid entirely, because it reads as a mistake rather than as a choice. Either the camera has genuinely moved somewhere else or it has not moved at all.
Which conveniently also avoids the jump cut, since a ninety degree change is far past the thirty degree rule that governs whether a cut reads as a cut.
Where it fails
- Anything with continuous player-controlled movement. The style needs a fixed camera, so it belongs to cutscenes, transitions and set pieces rather than to gameplay.
- Scenes that need spatial confusion. Planimetric staging is exceptionally legible, which is wrong for a scene about being lost or panicked.
- Long scenes. The rhythm is a strong flavor, and it wears at length. It works best in short beats.
- Doing it approximately. Eighty-eight degrees is worse than forty-five. The symmetry either is or is not, and a near miss reads as an error rather than as a softer version of the style.
Rules of thumb
- Camera perpendicular to a wall, in ninety degree increments. No three-quarter angles.
- Move along an axis or cut. Never arc.
- Hold. The hold is the style, and it is the part that gets cut for time.
- Cut ninety or a hundred and eighty degrees. A small angle change reads as a mistake.
- Snap the props too. The discipline is visible in the plan before it is visible in the shot.
- Exact or not at all. Approximately symmetrical is worse than deliberately asymmetrical.
One email when something new goes up. No newsletter, no schedule, nothing else.
Double opt-in, so watch for a confirmation email. Unsubscribe any time.