Plan a conversation shoot without confusing screen direction

Produce the complete planning deliverables from this source. Do not claim actual footage or a rehearsal.

FICTIONAL SEATED CONVERSATION GEOMETRY
Mathematical top-down coordinates: x increases rightward, y increases upward. Coordinate units are arbitrary planning units, not measured room dimensions. Seated actors A=(-1,0), B=(1,0); tabletop card initially centered at (0,0). The line through A and B is y=0.
OWNER-SUPPLIED RULE: every used camera must remain at y<0; A must stay screen-left of B in wide shots. Do not substitute a generic filming rule for this explicit requirement.
Candidates: C1=(0,-3), C2=(-2,-2), C3=(2,-2). Proposed C4=(0,3) is deliberately invalid under the rule. Reject it. A candidate lying exactly on y=0 would also fail the strict rule.
Required actions: R1 A giving B a card; R2 B reading it; R3 both reacting. The card begins on the tabletop at (0,0), so A must pick it up before the proposed transfer; preserve that initial state. Actors keep their seats. No unstated route around the axis, moving camera, elevated position, stunt or unsafe action is allowed. Room clearance, lens, camera height and actual shot contents are not supplied.

DELIVERABLE
Supply a simple actual top-down SVG map with the axis, actor/card positions, every candidate and clear rejected status for C4; a shot list with camera, aim and framing; an action-to-shot table; planned eyeline notes; unknowns; and a separate unperformed tabletop rehearsal/filming review. A wide must be proposed to show both actors and the entire transfer/reaction; singles may support but cannot prove both are visible. Use source IDs. Preserve exact source positions. No scripts, external files, remote fonts or photographic mockups. This is proposed geometry, not filmed footage or proof of screen direction.

Audit the map and shot list against the original sample coordinates and strict y<0 rule. Recompute camera admissibility and ideal left/right relations, and trace every required action to complete proposed coverage. Distinguish single-reaction support from both actors reacting in one wide. Report real lens/framing/rehearsal unknowns. Return complete corrected files and ledgers if anything disagrees.

Create a camera-side and action-coverage proposal using only MY INPUT. Return a self-contained safe SVG map or complete source with an explicit coordinate transform; classify every candidate by my rule; give camera/aim/framing shot rows, action-to-shot coverage, planned eyelines and unknowns. Separate complete coverage from partial single-angle support. Do not claim any shot exists.

MY INPUT
Actor and prop IDs, coordinates and initial states: [replace]
Coordinate units, origin and axis directions: [replace]
Owner-supplied action axis, allowed camera-side rule and required screen order: [replace]
Candidate camera positions and allowed movements: [replace]
Required actions and whether they must be visible together in one shot: [replace]
Excluded unsafe actions and physical access limits: [replace]
Known lens/framing/height information, or explicit unknowns: [replace]

Ask for missing essential facts or contradictory rules. Do not copy sample counts, positions or the sample's side rule. Do not invent footage, hidden room clearance or a rehearsal. Check every drawn coordinate and each action's coverage; explain any ideal projection assumptions. Provide a blank physical/viewpoint check for a person using safe positions, and keep it unperformed. Save originals unchanged; do not film, contact people or publish. If a mismatch is found, return complete corrected map and ledger.