subject_definitions:
<Subject 1> is Female Lead 1, the same young adult Chinese woman shown in <Picture 1> and <Picture 2>, with exact facial identity reinforced by <Picture 3>, <Picture 4>, and <Picture 5>. Preserve her softly tapered oval face, wide dark-brown almond eyes, delicate straight nose, full rose-colored lips, center-parted black hair with fine loose strands, elaborate antique-gold floral headpiece, red-amber-blue hanging beads, pale taupe-grey silk layers, and dense antique-gold embroidery. She has no beauty mark beneath either eye. Her fixed voice is an adult Chinese female low mezzo-contralto with strong chest resonance, dense dry warmth, slight restrained huskiness, no high-frequency sharpness, long breath support, and downward-resolving phrase endings; controlled, weighty, weathered, never sweet, girlish, breathy, shrill, shouted, or imitative of a male voice. She is (S1) in this clip.
<Picture 1> is the opening storyboard composition for [Shot 1], defining <Subject 1>'s three-quarter close-up, initial off-camera eyeline, pale embroidered costume, snowy woodland background, and shallow depth of field.
<Picture 2> is the target composition for the end of [Shot 1], defining <Subject 1>'s frontal close facial scale, direct lens gaze, costume crop, lighting, and background softness.

summary:
[keyframe completion + reference generation] Generate one continuous 5.0-second close-up of <Subject 1>. Begin from <Picture 1>, deliver the complete third dialogue block, and use a slow rightward camera truck, subtle arc, and gentle push-in to settle on the frontal composition of <Picture 2>. The camera transition is continuous, not a cut or dissolve. Only <Subject 1>'s clean dialogue is audible.

retention_analysis:
<Subject 1> (appears throughout [Shot 1]): fully_preserved - retain her exact face, no beauty mark, hairline, headpiece, hanging beads, pale taupe-grey and antique-gold costume, adult proportions, fixed low voice, and controlled expression throughout the changing viewpoint.
<Picture 1> ([Shot 1] opening composition): fully_preserved - retain the initial three-quarter face, off-camera eyeline, close crop, winter light and snowy background as the camera's starting state.
<Picture 2> ([Shot 1] ending composition): fully_preserved - use its frontal facial scale, direct gaze, costume framing, lighting and shallow background as the camera's final state.

detailed_description:
Preserve the exact live-action appearance of the storyboard images—cool winter shadows, natural softly lit skin, fine hair, and believable antique-gold embroidery—without redesigning or intensifying the visual style.

[Shot 1] Begin from the close composition of <Picture 1>. <Subject 1> is alone in frame, with Female Lead 2 remaining off-screen. Her head is held in restrained three-quarter view and her attention initially remains toward the unseen woman. The camera trucks slowly to its own right with small amplitude, adds a very subtle rightward arc around her face, and pushes closer at an even speed across the five-second shot. The viewpoint changes because the camera moves around her; <Subject 1> does not actively turn her head, and her face must not morph, recast, or duplicate. Keep the same snowy woodland, exposure, costume, ornaments and shallow depth of field. Her hands and weapon stay outside the crop. There is no cut, dissolve, subtitle, caption, fight gesture or visible second person.

The three identity portraits remain facial references only. They must not replace either storyboard composition, introduce extra cuts, or change the approved costume, headpiece, snowfield, head angle, or direct-gaze destination.

From 0.00 to 1.98 seconds, <Subject 1> (S1) begins immediately in her fixed low mezzo-contralto, using dense chest resonance and restrained dry warmth: <d>[Chinese] 你的刀太锐。</d> Give the weight to “太锐” without becoming louder, harsh, or accusatory. Her lips close at 1.98 seconds.

From 1.98 to 2.08 seconds, hold a very brief silent pause. Her eyes lower only slightly, without a full head dip. From 2.08 to 4.70 seconds, <Subject 1> (S1) raises her gaze and continues in the exact same voice: <d>[Chinese] 得在鞘里好好藏藏。</d> Deliver it as calm recognition and warning held inside composure. The second “藏” is lower and steadier than the first. Every syllable must finish before her lips close.

Throughout the line, the camera continues its rightward truck, subtle arc, and gentle push. During the final second, decelerate and converge on <Picture 2>. By 4.70 seconds, her face is frontal and her gaze meets the lens without a sudden eye snap. From 4.70 to 5.00 seconds, hold the exact face scale, direct gaze, closed lips, costume crop and background softness of <Picture 2>. Do not repeat the line, add a new gesture, touch a weapon, or continue into action.

overall_soundscape:
Only the clean, dry Mandarin dialogue of <Subject 1> (S1) is audible. The brief pause and final hold are absolute silence. No wind, snow, breathing, fabric, beads, footsteps, weapon sound, room tone, reverberant ambience, or other sound effect is present.

non_diegetic_music:
N/A
