FORMAT: vertical 9:16, approximately 22 seconds. Use the FIRST reference video as the primary visual

FORMAT: vertical 9:16, approximately 22 seconds. Use the FIRST reference video as the primary visual reference for broadcast aesthetic, camera quality, atmosphere, and editing. Maximum realism is the absolute priority. CORE STORY A young Japanese woman casually drinks an alcoholic beverage and visibly finishes the entire glass. She later drives a small white hatchback at night. While moving along an ordinary urban road, the car drifts toward the edge, the front passenger-side wheel climbs a raised curb, and the vehicle jolts with believable suspension movement. The driver brakes and narrowly misses a pedestrian standing on the sidewalk. A Japanese female reporter ends the video with a serious warning about drunk driving. REALISM FIRST Every frame must resemble live-action footage from a real camera, never a glossy commercial, music video, action film, video game, or 3D animation. Prefer ordinary, slightly imperfect television footage over cinematic polish. Skin needs natural pores, fine texture, small imperfections, realistic highlights, and consistent facial features. Hair, eyes, teeth, hands, fingers, clothing, glassware, vehicle parts, road markings, shadows, and reflections must remain coherent from frame to frame. Avoid beauty filters, waxy skin, plastic faces, artificial symmetry, excessive sharpening, oversaturation, theatrical lighting, or unnaturally clean surfaces. REAL-WORLD MOTION All motion must follow anatomy, gravity, inertia, friction, momentum, and believable reaction time. Actions progress continuously: no teleporting, sudden pose changes, frozen limbs, robotic gestures, rubbery bodies, sliding feet, floating objects, or motion without cause. Each action has a visible beginning, transition, and ending. Reaching involves natural arm and body shifts; hands maintain contact with objects; objects respond to gravity and applied forces. Expressions develop in response to events rather than appearing exaggerated or preset. Keep movement understated and documentary-like. CHARACTER CONTINUITY DRIVER: the same young adult Japanese woman in every relevant shot, with dark shoulder-length hair and bangs, wearing a cream ribbed long-sleeve top. Preserve her face, hairstyle, clothing, age, body proportions, and skin tone. At first she is casual and overconfident; while driving she becomes uncertain; after the near miss she looks genuinely shaken but does not perform exaggerated theatrical panic. PASSENGER: an older Japanese adult with short dark hair, glasses, and a blue-purple top. Keep appearance, glasses, clothing, and proportions consistent wherever visible. REPORTER: a Japanese female reporter wearing a dark navy suit and holding a realistic broadcast microphone. She looks into the camera with a serious, professional, restrained expression. Do not swap identities, change outfits between connected shots, duplicate people, morph faces, or alter body proportions. SCENE 1 — DRINKING, APPROXIMATELY 0–6 SECONDS Interior: a modest Japanese restaurant or dining room with warm practical lighting, ordinary furniture, a table, and realistic tableware. It should feel lived-in, not like a commercial set. Frame the woman at a natural medium-close distance, with imperfect composition typical of television news footage. She sits naturally at the table beside a transparent glass filled with an alcoholic beverage. Her posture is relaxed. She looks casually confident and says in Japanese: 「これくらいなら、全然平気。」 Her mouth movement must correspond to the Japanese dialogue. Use natural blinking, eye movements, breathing, and restrained facial motion. FULL-GLASS DRINKING ACTION — ESSENTIAL Show the entire drinking action continuously, without cutting away before she finishes. The glass begins full and visibly ends empty. 1. Her hand approaches the glass in a natural path. 2. Her fingers close around it with correct anatomy and believable grip. 3. She lifts it using coordinated wrist, elbow, shoulder, and hand movement. 4. She brings the rim to her lips without the glass passing through her face. 5. She tilts the glass gradually and drinks continuously. 6. Include subtle swallowing motions and believable movement around her throat and jaw. 7. The visible liquid level decreases consistently as she drinks. 8. She consumes the remaining liquid, including the final amount. 9. She lowers the glass after finishing. 10. The glass is visibly empty when lowered or placed on the table. Do not make liquid disappear instantly. Do not refill the glass between frames. Do not change its shape, size, color, or position without a physical movement. Do not hide the end of the action behind an edit, hand, subtitle, or camera cut. No impossible liquid flow, liquid passing through glass, fused fingers, or glass teleportation. Her head and wrist must tilt in a coordinated and believable way. Keep the drinking ordinary and unforced. TRANSITION TO THE CAR, APPROXIMATELY 6–8 SECONDS Use a straightforward documentary cut to the same woman driving the same small white hatchback. Her face, bangs, hair, and cream ribbed top must match. The interior needs plausible dashboard geometry, steering wheel, seats, windows, materials, and reflections. Her hands and arms must relate naturally to the wheel and seat. No teleportation or impossible camera placement. Keep the vehicle design consistent in all shots. SCENE 2 — MOVING THROUGH THE CITY AT NIGHT, APPROXIMATELY 8–13 SECONDS The car travels forward along an ordinary Japanese urban road at night. The car must visibly move: road markings, buildings, and lights pass relative to it, and wheels rotate with speed. It must not look parked while only the camera moves. Use a plausible urban speed and a consistent road layout. Include ordinary streetlights, sidewalks, buildings, lane markings, and slightly wet asphalt with restrained, physically correct reflections. The woman looks toward the road, not continuously toward the camera. Her hands remain in believable contact with the steering wheel. Her shoulders and torso respond subtly to steering and vehicle motion. She says in Japanese: 「はい、飲みました。」 Make the speech natural, with realistic timing and lip synchronization. Do not accelerate the voice unnaturally to force the sentence into the shot. VISIBLE IMPAIRED DRIVING — CLEARLY DRUNK, APPROXIMATELY 13–16 SECONDS The woman must visibly drive like someone impaired by alcohol, not like a normal careful driver. Show believable impairment without turning it into comedy: her gaze is slightly unfocused, her attention drifts, her steering corrections arrive late, and she makes a small uneven correction after the car begins wandering toward the road edge. Her upper body moves subtly with the car; she grips the wheel too tightly, then adjusts her hands. She briefly looks uncertain about the road position and reacts a fraction too late. Do not make her wildly flail, close her eyes for a long time, or behave like a cartoon. The impairment should be visible through delayed responses, poor lane discipline, inconsistent steering, and a hesitant, unstable trajectory. The car gradually wanders within its lane, then drifts toward the right edge. It first makes a small correction, overcorrects slightly, and moves toward the raised curb. Each steering input must cause a corresponding change in the car's path with realistic delay and inertia. The vehicle must remain in motion; no sudden sideways teleport or unexplained direction change. She says in Japanese: 「大丈夫、大丈夫……」 Her words sound like an attempt to reassure herself, with slightly uncertain delivery. Her expression shifts from confidence to concern. The passenger, if visible, may glance toward her or the road with concern, but should not distract from the driving action. Keep the curb, sidewalk, pedestrian, road, and buildings spatially consistent. Do not change the side of the road or relocate the pedestrian between cuts. CURB CONTACT AND SUDDEN CORRECTION, APPROXIMATELY 16–18 SECONDS The car must not gently glide onto the curb and stop. The moment feels sudden and dangerous. Its front passenger-side wheel approaches the raised curb at a believable angle and speed. The driver notices late and makes an imperfect steering correction while braking. The tire strikes the curb edge, climbs onto it, and causes a distinct but realistic jolt through the wheel, suspension, and body. The car body rocks briefly; the suspension compresses on the curb side and rebounds naturally. The tire remains in contact with the road/curb surface and does not pass through it. The impact is noticeable in the camera framing and vehicle movement, but it is not an explosive crash. Clearly show the moving car approach the curb, tire contact and climb, body jolt, late driver reaction, and slight lurch toward the sidewalk before sharp braking. The car must not leap, float, spin, roll over, or deform. The movement must be abrupt enough to communicate loss of control, but remain physically plausible for a small hatchback hitting a curb at low urban speed. Use a short natural camera shake caused by the operator reacting to the sudden event, not artificial action-movie shake. No sparks, explosions, dramatic crash effects, or slow motion. PEDESTRIAN NEAR MISS, APPROXIMATELY 18–20 SECONDS A pedestrian stands on the sidewalk close to the curb, clearly outside the main roadway. The car approaches dangerously close and briefly intrudes onto the curb edge, but never touches the pedestrian. The pedestrian notices the vehicle, turns, recoils, and takes one or two quick steps backward, moving their arms naturally to regain balance. React only after the vehicle becomes threatening. The driver reacts late, then brakes hard. The front end dips slightly under braking, the suspension responds, and the tires slow the vehicle with realistic deceleration. The car stops or slows just before contact, leaving a clearly visible gap. The near miss should feel tense because the car comes uncomfortably close and its motion is unstable—not because of a collision. No contact, injury, or impact with the pedestrian. The car must not drive down the sidewalk or through the pedestrian’s position. The woman exclaims: 「あっ、危ない！」 After the car slows or stops, she says with restrained shock: 「うそ……危なかった……」 Her face and breathing show a sudden rush of fear, with natural widened eyes and tense posture. Avoid cartoon panic, screaming theatrically, or exaggerated gestures. SCENE 3 — REPORTER, APPROXIMATELY 20–22 SECONDS Cut to a Japanese female reporter in a dark navy suit holding a professional handheld microphone. She stands in a believable nighttime reporting location and looks directly into the camera. The shot must resemble an actual Japanese television news segment, with ordinary broadcast-camera quality and practical lighting. She speaks seriously and clearly in Japanese: 「飲酒運転は、命を危険にさらします。絶対にやめましょう。」 Synchronize her natural mouth and jaw movements with the Japanese speech. Include breathing and blinking. Keep the microphone correctly held and her expression serious; no forced smile or cinematic posing. JAPANESE NEWS GRAPHICS — RESTORE THEM Do not remove the television-news captions and banners. Add authentic Japanese TV news graphics across the relevant scenes, inspired by the FIRST reference video. Use a compact red-and-white headline strip or blue-and-white information bar, Japanese subtitles for spoken lines, and a small restrained news label where appropriate. Use high-contrast stable lettering, with short headlines such as 「飲酒運転の危険」 or 「危険な運転に注意」 and subtitles matching dialogue. Avoid garbled, flickering, or changing text. Do not cover the glass, drinking action, curb contact, pedestrian, or reporter’s face. Keep banners in safe areas and subtitles near the bottom. Use red-and-white warning graphics during driving and a red lower-third during the reporter’s warning. ABSOLUTELY NO DATES, CLOCKS, TIMESTAMPS, TIME CODES, DIGITAL COUNTERS, OR NUMERICAL TIME OVERLAYS anywhere in the video, including corners, banners, driving footage, and reporter shots. News banners and Japanese subtitles ARE required; only time-related overlays are forbidden. Do not remove all graphics to satisfy the no-time requirement. CAMERA AND EDITING Use documentary editing and natural handheld framing: subtle operator movement, imperfect framing, believable focus adjustments, natural motion blur, realistic exposure changes, and modest low-light sensor grain. This is ordinary Japanese broadcast footage, not cinema. Cut only when the location or shot changes; never interrupt the continuous drinking action. Preserve the order: drinking, driving, drift, curb contact, near miss, reporter warning. Avoid impossible camera positions, excessive zooms, whip pans, lens flare, slow motion, and glossy grading. SOUND Use natural Japanese voices and accurate lip sync. The woman sounds casual, then uncertain and alarmed; the reporter is serious. Include restaurant, glass, engine, tire, night, curb-bump, and braking sounds, synchronized to action. No dramatic music or excessive effects. TEMPORAL AND OBJECT CONSISTENCY Maintain consistent faces, hair, clothing, proportions, car, glass, road layout, lighting, and environment. Liquid only decreases while drinking; the glass stays empty afterward. The hatchback’s body, windows, lights, wheels, and interior remain unchanged. Wheels rotate with motion; shadows and reflections match their light sources. No flickering, morphing, duplication, sudden position jumps, or impossible speed changes. All actions obey gravity, inertia, friction, and reaction time. STRICT NEGATIVE REQUIREMENTS No CGI or game-like visuals; plastic skin; facial morphing; distorted or extra fingers; robotic movement; sliding feet; floating or teleporting objects; changing clothing or vehicle shape; incorrect wheel rotation; floating cars; impossible steering, braking, or suspension; tire passing through curb; liquid disappearing or refilling; pedestrian teleportation; collision or injury; cartoon panic; excessive camera shake; fake reflections; oversaturated cinematic grading; slow motion; futuristic graphics; illegible or flickering Japanese text; dates, clocks, timestamps, timecodes, or digital counters. Do not remove news banners or subtitles: only time-related overlays are forbidden. FINAL DIRECTIVE Generate one coherent 22-second vertical 9:16 video. Use the FIRST reference clip for image quality, Japanese broadcast graphics, and documentary style. Preserve the story and characters. The woman drinks the entire glass continuously; the moving hatchback drifts, strikes and climbs the curb with a distinct realistic jolt, and narrowly misses the pedestrian without contact. End with the reporter warning against drunk driving. Every movement must look real. Realism over spectacle; authentic news footage over cinematic polish.