Ежедневные бонусы и бонус за подписку — после регистрации
Промпт
[TASK} New photorealistic cinematic video of the people from @image1 and @image2. @video1 is reference ONLY for camera, framing, movement, shot order, edit rhythm, actions, gestures, gaze, emotional beats, singing rhythm, staging. NOT a face swap, overlay, reenactment or edit of the original actors. Build characters from the PHOTOS, then have them perform the video's actions. On any anatomy conflict, the photo wins. [CHARACTERS} @image1: driver, SCREEN-RIGHT in opening two-shot, sings. @image2: passenger, SCREEN-LEFT, holds small fluffy dog, reacts. Viewer's-image positions. No mirroring, seat or identity swaps. [FACE — CRITICAL] Full head and face from each photo: head size, skull, face proportions, forehead, hairline, eye shape/size/spacing/lids, brows, nose bridge/width/tip/nostrils/projection, lip thickness, mouth width, lip contours, cheeks, cheekbones, jaw, chin, ears, skin tone, age, distinguishing features. Never keep source actors' nose, mouth, lips, jaw, cheeks, head proportions; no stretching face over another structure, no blending. Small nose, narrow mouth, thin lips or different jaw stay intact in every expression and angle; only natural deformation from singing, smiling, surprise. [BODY & OUTFIT] From photos: gender presentation, physique, shoulders, neck, proportions, hair, clothing, accessories. No source physique under a new face. Fit each person naturally into the seat; adapt gestures to their proportions, no stretching to source outlines. No source glasses, jackets, jewelry, hair, makeup unless in photos. Unseen details: restrained, consistent completion. [PERFORMANCE} Transfer meaning, timing, intensity of expressions, not geometry, through each person's own features. Same identity frontal, 3/4, profile; no drift in wide mouth openings, smiles, head turns, hand-to-face gestures. [FORMAT} Source horizontal ratio, real duration (~18s), source shot order and cut points; timings below are guides. No added shots, slow motion, zooms, camera moves. [SETTING} Night, open-top car, dark upholstery, blurred city lights, soft cinematic light on faces. Source vehicle layout, steering-wheel side, camera placement. New coherent scene; no source pixels or facial outlines. Dog matches source look, scale, role; realistic hand-dog contact. [DRIVER} Actively sings along to @audio1, not speech, silent smiling or random mouth movement. Lip closures on consonants, jaw/lip shapes per syllable, natural vowels on held notes, pauses on time; cheeks, eyes, brows engaged. No enlarged mouth/lips, no nose/jaw/cheek change. Playful confidence, smiles, head moves, glances to passenger, later finger gesture near face on time. Only visible singer. [PASSENGER} Restrained disbelief, awkward surprise, confusion, source timing and gaze. No lip-sync, new gestures, dialogue. [SHOTS} 1 (~0:00-0:04): frontal medium two-shot into the car; driver sings, follows source steering and head moves; passenger holds dog, looks at driver in disbelief. 2 (~0:04-0:07): passenger close-up, full @image2 geometry; confused reaction, dog where visible; singing offscreen. 3 (~0:07-0:16): driver 3/4/profile close-up, main singing shot; full @image1 profile, nose projection, lips, jaw; phrasing, smiles, gaze changes, face gesture; hands occlude without deforming features. No static portrait, mouth-flapping loop, source anatomy returning. 4 (~0:16-0:18): passenger close-up; hold stunned, questioning look; dog continuity, source ending; no extra action. [QUALITY} Natural skin texture, coherent anatomy, real fabric, believable reflections. Match source camera language, DOF, lighting, motion blur. No mask edges, halos, head-neck mismatch, frozen expressions, waxy skin, warped backgrounds, text, subtitles, watermarks. [AUDIO} @audio1 (18s) is the soundtrack: play it unchanged 0:00-0:18, in sync second-for-second. No generated voice or music; never replace, restart, stretch, loop or rearrange it. Driver lip-syncs to @audio1 vocals. No dialogue, narration, second singer, effects.
Описание
В кадре двое едут в машине: водитель поёт, а пассажир с пушистой собачкой реагирует на его исполнение и подхватывает настроение сцены. Загрузите фотографии обоих участников — нейросеть сохранит их лица и создаст комедийный ролик с похожими жестами, взглядами и автомобильной атмосферой.
Вопросы
Сколько стоит?
Как обычная генерация видео в Seedance 2.5 на нашем сайте — точная цена показана на кнопке перед запуском. Если нейросеть не справилась, нейроны вернутся автоматически.
Нужно ли своё фото?
Да — загрузите одно чёткое фото лица анфас, без очков и фильтров. Нейросеть сохранит черты.
Можно изменить промпт?
Да. Нажмите «Создать такое же» — на открывшейся странице промпт можно поправить перед запуском: одежду, фон, надпись, цвет. Или скопируйте промпт и вставьте в генератор.
Где найти результат?
В разделе «Мои генерации»: оттуда можно скачать фото или сделать ещё вариант.