shot-specifier

Shot Specifier

Turn an approved storyboard unit into camera-executable shots. Do not add new story claims.

Required fields per shot

• shot ID and time range;
• semantic job: one dominant thing this shot must make visible;
• subject and environment;
• start composition and spatial state;
• blocking/action direction;
• camera scale, angle and movement only when purposeful;
• visible cause -> transformation -> result;
• end composition/state;
• handoff to the next shot;
• acceptance evidence and forbidden failure modes.

Camera discipline

Choose the simplest camera grammar that reveals the intended relation.

• Establish space before asking the viewer to compare parts of it.
• Use push-in when detail/depth is the new information.
• Use pull-back when the larger distribution/context is the new information.
• Use orbit/rotation when viewpoint itself is the concept.
• Use lateral move when comparing neighboring regions in one stable space.
• Use motivated cut when the narration changes conceptual layer or example.
• Do not use movement as decoration. A moving camera with no information gain is noise.

Blocking and causality

Resolve complex motion as: subject -> start position -> direction -> interaction/connection -> visible result -> end position/state.

Do not hide missing mechanics behind “then”, “suddenly”, “transforms” or “becomes”. If a model cannot physically execute the chain in one shot, split it.

Start/end frame strategy

Use a locked first frame when a shot must inherit exact space/identity/composition. Use a locked last frame only when the destination state is important enough to justify constraining motion. Do not force FL2V on every shot; over-anchoring can produce mechanical interpolation.

Explanatory-film rule

A shot is accepted only if a muted viewer could point to the visible evidence that corresponds to the narration. Beautiful atmosphere without explanatory evidence fails.