ee59efb3-29d3-4724-b6f7-92d05013555b

@Daniel delivers the closer to off-camera @Marcus: "Help me find her and I'll make sure your name reaches the right desk. Men get promoted for this kind of cooperation." Delivery: warmer than the lie in 3.07 — this is a favor being offered, not a command. Mid-volume, pace approximately 3.5 words per second. A small smile sits at the corners of his mouth during delivery but does NOT reach his eyes — the warmth is performed, not felt. Subtle chin dip lands on "promoted" as punctuation. Right shoulder lifts a fraction on "cooperation" — small emphasis gesture. Eyes locked on @Marcus throughout, no darting. Single blink between the two sentences. Final beat: mouth closes, the small smile holds half a second then softens to neutral. Single controlled blink in the last quarter-second. Chin stays in lowered position. Silence. Camera: locked, organic micro-drift. Focus on eyes. Warm key constant. Real time, closing-the-deal weight. The shattered polygon/glass segmentation visible in the reference image is only an image composition artifact and must not appear in the generated video. Reconstruct the character as a single, fully intact, photorealistic human face with continuous skin, natural facial anatomy, and seamless textures. Remove all white gaps, polygon divisions, cracks, broken-glass fragments, cut lines, masks, overlays, or geometric segmentation. Treat the fragmented reference as if all pieces belong to one complete face. The final output must show a completely normal, realistic person with an uninterrupted face while preserving the overall hairstyle, facial proportions, expression, clothing, lighting, and camera framing from the reference..