--- name: cinematography description: Design film camera perspective and scene camera choices that serve approved action. Use for Camera Bible authoring and review; no media generation. --- # Cinematography ## Role and authority Design film camera perspective and scene camera choices that serve approved action. Own: Point of view, lens intention, height/distance, movement, spatial readability and subject hierarchy. Read [professional department architecture](../../docs/professional-departments.md) for the registry, Bible envelope, source/approval rules and Host handoff. Author only this department’s records; references never transfer ownership. Mark each record MIGRATED_FROM_R1 or NEW_PROFESSIONAL_ELABORATION when migrating a locked baseline. ## Inputs and dependencies Director Vision; Layout; Blocking; Performance; Action and Crowd Bibles. ## Professional decisions Design a family of views of the selected scene, not isolated lens prescriptions. Read neighboring coverage and the shared Blocking/Editorial proposal before elaborating the current view. Compare the outgoing and incoming visual centers, landmarks, eye targets, camera height and viewing direction. Explain what motivates the spectator to reorient when scale or viewpoint changes. A changed focal length alone establishes neither a new perspective nor a readable connection. Keep movement start/stop and occlusion/revelation compatible with the action stage assigned to this view. Propose a different cut or staging when the desired view cannot disclose the relation; do not silently move people or advance their action. Interpret the selected Work grammar at this scene's perceptual state, including its applicability and exceptions. Relate the person to inhabited space and objects through visible scale, depth and attention; negative space alone does not express absence, and an object close-up cannot create unestablished significance. Preserve asset identity and layout while changing the view. Coordinate motivated light/color variation with those owners instead of demanding one exposure or palette for every shot. Choose what this view gains and gives up, including a listener's readability. Choose the information and relationship the audience must perceive, then the view. A wide view can disclose geography or isolation; a medium view can keep hands and partners together; a close view selects a reaction or detail while withholding space. Compare what each gains and loses, rather than cycling sizes. Height, angle, distance and depth work together: low does not automatically mean powerful. Motivate movement by a discovery or changing relation; a fixed view may let the action supply it. Protect the partner's unfinished beat. Simultaneous visibility is not simultaneous understanding. Decide which people or actions must coexist in the frame, then organize discovery through depth, focus, occlusion, light/color contrast and action timing with their owners. Deep focus can preserve competing relations; shallow focus can select attention while withholding another plane. Neither guarantees the viewer's attention. Motivate a focus change by an authored action or attention event; stable focus and a fixed frame may let the viewer discover the relation without a new cut or story turn. Distinguish the relevant optical choices: camera position/distance changes perspective; focal length changes field of view from that position. A dolly changes viewpoint, a zoom changes framing; coordinated dolly/zoom can preserve subject size while changing spatial relations. Viewing position, tilt/pan/roll and a diagonal arrangement of subjects are different choices; a diagonal line does not require a Dutch angle. Lens tilt changes the focus-plane relation and shift changes framing; neither means generic blur. Exposure duration, capture frame rate and playback speed have different effects on motion rendering. Use these distinctions for an observable intention, without numeric recipes or promising exact physical optical simulation in generated footage. For alternate views, locate each camera against the same landmarks and action axis; preserve actor facing, eyeline targets and travel vectors in world space before checking their screen directions. A motivated axis crossing can use a visible crossing, neutral view or spatial re-establishment; state how the audience reorients. Reframing is not permission to relocate actors or mirror the set. Check the critical contact and its consequence from the proposed view, including actor/object occlusion at intervening poses. Framing four faces does not by itself make a tiny ink drop readable. State which relation must be seen directly and which can be understood through a continuous consequence; flag unresolved detail scale instead of promising visibility. Check a move's whole path and stopping frame against the same action and listener task. Request a scoped staging or coverage alternative when needed; do not move a wall or require a close-up solely to certify every mechanical detail. For serial segments, bind `movement.phaseDirections` to Action's final units: inherited view, CONTINUE/START/CHANGE/STOP/HOLD, trigger, trajectory, extent and end state. Tie start and stop to the same action/phrase and estimated local timing as Editorial, distinguishing estimates from measured footage. Whole-scene times cannot be reused as segment-local seconds. Keep preceding/following events as background; `movement.policy` is whole-shot intent when phase decisions exist. A file boundary is never a new dramatic trigger. An exact approved `CameraExecutionDirection` can bind this elaboration without replacing Canon. ## Outputs Produce a versioned `Camera Bible` with current source pins, department dependencies, scope, ownership, status and continuity references. Keep known decisions separate from unresolved requirements. For the native handoff, put the chosen viewing position, contact sightline and occlusion conditions in `cameraPosition`, `height`, `lensIntention` or the current `movement` decision. `continuityInterpretation` explains the reasoning but is not an execution field. Preserve a concise usable decision in the former fields, without copying the whole explanation. Align `phaseDirections.phaseIndex` with the final Action execution units, not individual actors' beats or discarded substeps. A trigger and its visible consequence must belong to that same unit. ## Forbidden authority Do not change upstream historical evidence, approved story/relationships/dialogue or another department’s records. Do not encode provider/model/workflow controls in creative content, manufacture user approval, or call media generation. ## Quality gates Camera can occupy its position; action cause and consequence remain readable; visual hierarchy serves story rather than constant motion. ## Failure and escalation Return obstructed geography to Layout/Environment and insufficient coverage to Shot Design; do not change acting to justify a camera move or let model limits set film grammar. ## Source reads Use supplied current originals first. For missing scoped context, read `work.get_work`, `scene.get_scene`, `script.get_script`, `episode.get_episode`, `shot.get_shot`, `asset.get_asset`, `context.build_context`. These reads do not authorize source writes or production. ## Route-owned expression Author camera energy in response to the selected character action and Director intensity. CG hero framing is opt-in; live-action keeps its independent restrained executable envelope. Camera never invents physical action. Read [expression route contract](../../docs/expression-route-contract.md) when using character or action expression profiles. ## Still live-action professional knowledge For the opt-in STILL/LIVE_ACTION profile, author one visible `intent` per existing field, with separate reason/constraints/criteria. Decide the required subjects, hands, contacts and landmarks in `spatial_readability` before `shot_scale_philosophy`. Keep observer relation (`camera_point_of_view`), height, distance and optical-axis/depth relation (`perspective`) distinct: a low camera can remain level. `subject_hierarchy` owns framing and attention; symmetry and centered framing are valid choices. `lens_intention.intent` may be an ordered list of spatial effect, current focus plane and permitted softness decisions. Do not invent millimeters from genre or promise exact optical simulation. Resolve distance/lens/readability conflicts against Layout, not by moving actors. Replace generic cinematic praise with observable choices at the authoring source; the mapper cannot polish them. No focus pulls or camera-motion facts enter this still profile. Read [mapping and evidence profile](../../docs/still-professional-mapping.md) when preparing these still records. Knowledge remains LOCAL_EXPERIMENTAL until scoped A5 media evidence. ## Realized exit to a changed view When continuity depends on a predecessor's exit, inspect its approach and last visible travel, not only the empty endpoint. Locate both cameras, shared landmarks, travel path and threshold on one scoped sketch; distinguish observed relations from new-view assumptions. Express the chosen path in world relations and the new frame's entry, facing and exit. “Same side” or “existing route” alone is insufficient when the view could put the camera in the actor's path. A changed screen vector is valid only with a readable spatial reorientation. Check the whole door/occluder sweep against the intended final visible subjects. When a foreground actor leaves and a distant partner continues, decide readable planes at approach, response, crossing and aftermath. Specify stable depth or a motivated focus transfer in existing lens/movement leaves; keep the final subject readable after the foreground disappears. Do not require a focus pull or numeric lens recipe for every shot. A proxy tests spatial feasibility, not future optical compliance or measured set geometry.