The camera-move vocabulary — named moves, 2.5D implementations
The classic move names, each with: what it physically is, what it says, and
the EXACT implementation in this skill's rig convention (one perspective
wrapper → one transform on the space → planes at translateZ). Camera pose
convention: transform: translate3d(−x, −y, z) rotateZ(−rz) rotateY(ry) rotateX(rx) on the space; scale(s) on the OUTER wrapper is the zoom axis
(post-projection = FOV change).
The two distinctions that only exist once you have depth planes:
- Pan/tilt (rotation) vs truck/pedestal (translation): rotation produces NO parallax; translation separates the planes. In 2.5D, big "pans across a scene" are almost always trucks.
- Zoom (outer scale) vs dolly (camera z): zoom magnifies everything equally; a dolly changes the relationships — near planes grow faster. This difference is the whole reason to build in 2.5D.
| Move | Physical | Says | 2.5D implementation | Watch out |
|---|---|---|---|---|
| Truck | camera slides left/right | follow travel, scan a scene, scene-to-scene transit | animate x; planes parallax automatically |
the workhorse — preset A's glides are trucks |
| Pedestal | camera rises/falls, axis level | change of level, hands→faces reveals | animate y |
pairs with truck for L-shaped reveals |
| Dolly in/out | camera moves toward/away | entering the world / leaving it; depth context | animate z (+ = in); planes separate |
never a bare scale() — that's zoom, and the anti-AI-camera rule wants dollies paired with a lateral or roll component |
| Zoom in/out | focal length change | attention squeeze without spatial change | animate outer-wrapper scale 1→~1.2 |
flat feel is the POINT of zoom; use when you don't want parallax (archival/CCTV/broadcast flavors) |
| Dolly-zoom (Vertigo) | dolly one way, zoom the other | unease, realization, world-shift | z: 0→c while outer scale compensates s = (P − c)/P (P = perspective, 1100): focal plane holds size, background crawls |
the strongest move in the kit — one per piece, on the biggest narrative turn, tier-5 territory |
| Pan | camera rotates on Y | glance left/right, tension between two subjects | animate ry, SMALL (≤ ~6°) |
flat planes go edge-on at big angles and the illusion dies; big scans = truck instead |
| Tilt | camera rotates on X | reveal up/down (knife → face) | animate rx, SMALL (≤ ~5°) |
perspective distorts type — never tilt across crisp focal copy (A5 screen-flat plane) |
| Roll | camera rotates on Z | disorientation, energy accents | animate rz |
already rationed by the grammar: ±0.9° poses, hit accents |
| Crane / jib | camera arcs up + tilts down (or reverse) | scale, arrival, finale | path: y down→up eased WITH compensating rx so the focal stays framed |
the classic closer; pair with a section seam |
| Whip (pan/truck) | violently fast lateral | hard transition, chaos accent | truck at minimum T (~10–14f) + velocity smear per plane; landing sharp; can hide a cut at peak smear | this is a TRANSITION, not a shot; A14 ledger — twice max |
| Handheld | operator's body noise | authenticity, urgency, doc energy | preset B (Reactive Shake), scoped | field-judged: never the standing behavior |
| Steadicam / follow | smoothed pursuit of a subject | intimacy with a moving subject | preset C (Element-Follow) — camera home tracks the focal with inertia | roadmap; needs travelling focals |
| Rack focus | focus shifts between depths | redirect attention without moving | animate focusZ: crossfade per-plane DOF blurs (far crisp→soft while near soft→crisp) over 12–20f |
the "camera move" of attention; obeys CAM-2 — it must LAND on the narrative target |
| Signature / custom | the rope-swing | style identity | a named pose-path defined once in the style bible, reused as the piece's motif | CAM-1: vary its parameters between uses |
Composing them
- Real moves are pairs: truck+pan (scan with a look), dolly+roll (the anti-AI pairing), pedestal+tilt (crane), truck+zoom (push-slide). Single- axis moves read mechanical — same law as element arcs (A2.7).
- Dial gates the vocabulary: 1 → cuts only · 2 → zoom/rack focus only (no spatial motion) · 3 → truck/pedestal/dolly as motivated phrases · 4 → full kit incl. crane paths · 5 → continuous pose-graph composition.
- Music map places them: glides/trucks land on section seams (preset A); dolly-zoom and crane belong to the piece's ONE biggest moment; whips are hot-section transitions; rack focus rides vocal/narrative handoffs.
- Every move remains a phrase (CAM-4): wind-up, travel, settle-into-breath — and the A14 ledger applies to moves exactly as to element devices.
All implementations verified in the CamMoves glossary demo (fx-poc) — rendered, viewed, each move labeled.