subject_definitions:
<Subject 1> is the adult Chinese woman whose face and hair come from <Picture 1> and whose 170 cm body proportions and costume come from <Picture 2>: black low ponytail, small silver earrings, cream shallow-V fine-knit top, camel high-waisted plain trousers and a narrow dark-brown belt. The three views depict one woman; her face has natural matte skin texture.
<Subject 2> is the adult Chinese man whose face and hair come from <Picture 3> and whose 180 cm body proportions and costume come from <Picture 4>: short black hair, clean-shaven face, plain solid navy cotton button shirt with sleeves rolled to the forearms, gray-taupe straight trousers and a dark-brown belt. The three views depict one man; his face has natural matte skin texture.
<Subject 3> is the lived-in apartment in the storyboard images: a brown rectangular table, red closed notebook and blue-white mug at table-left, a clear daisy vase at center, sofa and bookshelf at room-left, television at room-right, rear dining/balcony opening, and a solid gray entrance door near screen-right. Beside that entrance is a narrow freestanding dark wooden three-drawer cabinet carrying a white plant pot and a small dark bowl. This cabinet is separate from the television console. Its layout is referenced from <Picture 5>, <Picture 6>.
<Picture 5> is a storyboard reference for [Shot 1], guiding the already open-arm standing pose at the opening. It supplies shot planning, not a compulsory first frame.
<Picture 6> is a storyboard reference for [Shot 1], guiding the later pointing woman and outward cloth gesture toward the entrance. It supplies shot planning, not a compulsory first frame.

summary:
[reference generation] Generate a 7.292-second live-action domestic argument segment with <Subject 1> and <Subject 2> in <Subject 3>, following <Picture 5>, <Picture 6> as sequential shot-planning references. Keep knee-to-head framing and furious screaming from the opening through the ending, with newly generated Mandarin dialogue, physical room sound and no music.

retention_analysis:
<Subject 1> (appears in [Shot 1]): fully_preserved - retain her identity, age, body proportions, hair and costume in every shot.
<Subject 2> (appears in [Shot 1]): fully_preserved - retain his identity, age, body proportions, hair and costume until his described departure.
<Subject 3> (appears in [Shot 1]): fully_preserved - retain the architecture, lighting direction, object identities and established room geography as actions change their states.
<Picture 5> ([Shot 1] storyboard plan): partially_preserved - retain the already open-arm standing pose at the opening; adapt exact limb positions and mouth shapes to the continuous actions and alternating spoken turns.
<Picture 6> ([Shot 1] storyboard plan): partially_preserved - retain the later pointing woman and outward cloth gesture toward the entrance; adapt exact limb positions and mouth shapes to the continuous actions and alternating spoken turns.

detailed_description:
Observational 16:9 live-action drama: neutral daylight, gray-white walls, warm wood, readable architectural depth, matte skin and irregular natural pores. Frame each person from the knees to the complete top of the head; gentle camera-height adjustments preserve this crop, headroom and natural proportions.
[Shot 1] Hold a standing eye-level two-shot in <Subject 3>, using <Picture 5> for the starting body arrangement. <Subject 1>, in cream knit, stands on the left with both arms already spread; <Subject 2>, in plain navy, stands on the right holding the pale cloth in his left fist. Both chairs remain pulled back. The framing includes their knees and complete heads.

Using an adult Mandarin female voice in a high register, piercing and sharp, with a dry hoarse grain and strained tearing edges; fiercely projected screams with clear consonants, uneven angry breath pressure and brief voice cracks, <Subject 1> (S1) immediately screams <d>[Chinese] 你有本事跟人去睡呀</d> directly into his face across the table. She converts the existing open-arm pose into one accusatory right index finger, while her left hand drops and clenches. Her brows stay tightly drawn, the jaw opens broadly on stressed vowels, and the pointed challenge cuts through a frayed breath. She strikes the daring phrase and the final sleeping accusation, completing the last syllable before he responds. The movement is a continuation of her stance rather than a fresh opening of both arms.

<Subject 2> listens with a closed mouth and tense cheek muscles. After a short hostile intake, <Subject 2> (S2) replies using an adult Mandarin male voice in a mid-low register, thick chest resonance and a coarse hard edge; forceful rough angry shouting with clipped consonants and hard breath attacks: <d>[Chinese] 对 方一诺 你们就一帮烂货</d> He punches out the opening agreement, gives the woman's full name a brief separate bite, then drives the insult outward in one coarse chest-supported shout. His gaze remains fixed on her; there is no amused grin. She holds the pointing hand steady, lips closed while listening, with furious eyes and shoulders rigid against the insult.

<Subject 2> (S2) takes a compact angry breath and bellows <d>[Chinese] 我就喜欢到外面睡</d>, stressing the preference and the outward destination. His cloth-holding left arm sweeps once toward the solid entrance door near screen-right, developing the gesture in <Picture 6>. The cloth trails naturally from his fist and remains held; the vase is not struck. His right hand stays near the table edge. The camera holds the same axis and crop, preserving the architectural background without pulling away from the actors.

Finish all words by about 6.75 seconds. During the remaining roughly half-second, hold his outward arm and her accusing finger in tense opposition while their breathing remains fierce. Both stay at the table; the door stays shut and the cloth remains in his left hand. The final syllable is fully voiced before his mouth closes into an angry set, leaving a clean action state for the next exchange.

Maintain full-intensity rage from the first word through the ending. Speak every quoted line once in order; listeners hold closed lips and furious eyes. Short angry inhalations preserve clear words.

overall_soundscape:
Low indoor room tone, short dry angry breaths, shirt and knit rustles, and a soft cloth swish form the physical sound bed. Keep voices present and unprocessed, with natural short room reflections and clear syllables.

non_diegetic_music:
N/A
