Фейк-коммершиал, шесть шагов

Как собрать рекламный ролик без съёмки

Фейк-коммершиал — это рекламный ролик для бренда, который тебя не нанимал. Делается в основном для портфолио. ОБЯЗАТЕЛЬНО выкладываем в Instagram.

6шагов
0дней съёмки
1человек в команде
30 сготового материала

Что понадобится

Четыре инструмента. Первые три — по подписке, монтажка бесплатная, но можно и воспользоваться монтажёром.

Сначала разберёмся со словами

Дальше по тексту будет несколько терминов. Сразу как встречаете непонятное слово — возвращайтесь к этому списку.

Референс
Картинка или видео, которое ты прикрепляешь к промпту как образец. Нейросеть смотрит на него и повторяет по нему.
Плейт
Опорная картинка, зафиксированная один раз: лицо, костюм или локация. К ней потом цепляются все остальные генерации, чтобы ничего не менялось.
Лок
От английского lock — замок. Это плейт лица, снятый на плоском сером фоне без всякого света. Нужен, чтобы нейросеть взяла оттуда только внешность, а свет задала уже по твоему промпту.
Дрейф
Когда персонаж плывёт от кадра к кадру. В первом кадре одна девушка, в десятом — уже как будто её двоюродная сестра.
Карточка персонажа
Картинка, где герой показан с разных ракурсов и уже в нужной одежде. Нужна не всегда — только когда генераций несколько и мелкие детали образа начинают расходиться.

Порядок здесь важнее промптов

Каждый следующий шаг опирается на результат предыдущего, и пропуск любого из них возвращается переделками. Лицо без опорной картинки (плейта) плывёт между кадрами. А мелкие детали образа — длину носков, посадку пояса, укладку — нейросеть додумывает заново на каждой генерации, пока они не зафиксированы карточкой персонажа.

Отдельно про свет: он должен быть решён один раз и держаться все тридцать секунд. Если референс несёт один свет, а промпт требует другой, нейросеть начинает спорить сама с собой, и свести такие кадры в один ролик на цветокоррекции уже не выйдет.

Промпты ниже — это всё же шаблон

Они написаны под конкретную задачу: теннис, спортивный бренд, одежда. Если делаешь ролик про другое — меняй описание внешности, вещей и локации, а структуру промпта и все технические блоки оставляй как есть. Именно они отвечают за то, чтобы картинка не разваливалась. Поменять промпт можно через ChatGPT или Claude.

01Лицо

Изображение лица в Soul Cinema

Сначала нужно лицо, которое не будет меняться от кадра к кадру. Soul на этом шаге работает как генератор вариантов: гоняешь 5–6 штук и выбираешь одно.

Важно, где происходит сцена. Мы начали с бетонной стены в холодном нейтральном свете и получили серию невыразительных лиц. Пляж, зонт и тёплый рассеянный свет — тот регистр, который Soul отрисовывает лучше всего. Красоту приходится прописывать словами так же явно, как цвет волос: костная структура, посадка глаз, форма губ.

Отобранное лицо из Soul Cinema
Отобранное лицо из Soul Cinema
Промпт изображения — Higgsfield Soul Cinema
A vertical medium close-up selfie of a beautiful Slavic woman, held at arm's length on a phone, slightly below eye level. Model-grade symmetrical features - high defined cheekbones, a clean narrow jawline, a long neck. Fair skin with a light warm sun-kissed glow, flushed across the cheekbones and nose. Long jet black hair with a cool blue-black cast, loose and slightly windblown, falling forward over one shoulder in a soft natural wave, middle part, fine flyaways lifting at the hairline. Large dark almond eyes with a slight upward tilt at the outer corner, long dense lashes. Straight groomed dark brows with a soft arch. A narrow straight nose. Full lips with a defined cupid's bow, natural color, lips closed with a relaxed easy half-smile, eyes directly into the lens. A small beauty mark below the right cheekbone. A fine silver chain at the collarbone. She wears a plain black thin-strap top.

Behind her, a tropical beach club - a weathered timber structure with a thatched roof and bamboo railing, pale coral stone stacked low beneath it, out-of-focus palm fronds and a bright hazy sky at the frame edge. Bright daylight filtered through a large white beach umbrella directly behind camera, wrapping her face in soft diffused light - open and even, no hard shadow across the face, no heavy nose shadow, no blown-out hotspot on the skin. Warm earthy midtones, gentle highlights, relaxed summer mood.

Front-facing phone lens, the face sharp and the background falling into mild natural bokeh. Real fine pore texture, peach fuzz at the jaw and hairline, subsurface scattering at the ear edges, hair rendered strand by strand, real fabric weave at the strap. Skin flawless, even and flattering - no acne, no blemishes, no rough or enlarged pores. Soft natural grain. Photographed not generated.
02Лок

Канонический плейт в Nano Banana Pro

Селфи — это фотография, а не референс. В нём уже «зашит» свет: тёплый бок, тень под скулой, размытый фон. Seedance тянет из референса всё сразу, вместе с лицом он потянет и этот свет — и подерётся с тем, что ты фиксируешь в сцене.

Плюс селфи снято одним ракурсом. Лица в профиль и сверху там нет, модель их додумает, и каждый раз по-своему. Это и есть дрейф, из-за которого героиня в первом кадре и в десятом выглядит как двоюродная сестра самой себя.

Лок — та же картинка, но на плоском сером фоне и без света — решает обе проблемы: нулевой свет означает, что свет целиком задаётся промптом сцены, а полная анатомия словами даёт модели, из чего собирать любой ракурс.

Канонический плейт, к нему цепляется всё дальнейшее
Канонический плейт, к нему цепляется всё дальнейшее

Прикрепить к промпту

  1. Отобранное селфи из шага 01
Промпт лока — Nano Banana Pro, 3:4
A clean cinema-character-reference 3:4 headshot of the same woman as the attached face plate, framed from the forehead down to the upper chest with the face filling most of the frame - a true close-up, not a portrait with space around it.

Lean athletic build with a long neck and defined collarbones. Fair-to-medium skin with a warm golden undertone and a soft matte finish, a natural flush across the cheekbones and the bridge of the nose. Oval face with a broad smooth forehead, flat temples, high sharply projecting cheekbones with a clear hollow beneath, a narrow tapering jaw with a soft clean angle, a rounded chin with gentle projection, a long uninterrupted line from ear to chin. Large almond eyes, wide-set and slightly deep-set, with a faint upward tilt at the outer corner, a visible clean lid crease, warm hazel-brown irises that lighten toward the pupil with fine amber flecks and a soft dark limbal ring, smooth flat under-eye structure. Thick dark brows, straight through the body with a low soft arch toward the tail, dense with visible individual hairs sweeping upward and outward at the head. Long dark lashes, dense and separated with a natural curl on the upper lid, shorter and finer on the lower. A narrow straight nose with a fine even bridge, a small softly rounded tip with slight projection, narrow nostrils barely visible from the front. Full lips with a slightly fuller lower lip, a clearly defined cupid's bow, a short shallow philtrum, a medium mouth width with softly downturned corners, natural warm rose lip color with fine vertical surface texture. Small close-set ears, mostly covered by the hair. Long dark hair, near-black with a warm dark brown cast in the depths, straight to softly waved with a natural bend through the lengths, swept back off the face at the crown and falling loose down both sides past the collarbone, a low soft hairline with fine baby hairs and flyaways lifting at the temples. Bare-skin makeup register - no visible base, a light natural lash and brow grooming, nothing on the lid, natural lip. A cluster of three small dark beauty marks on the left cheek below the outer corner of the eye, one small mark low on the right side of the neck above the collarbone.

She wears a plain black thin-strap camisole, no jewelry, no logos, no graphics. Body squared to camera, head level, neutral relaxed expression, eyes directly to camera, lips closed and relaxed, subtle controlled energy.

The background is a single flat 18% neutral gray field - one uniform value at every pixel corner to corner, identical directly beside the subject and in the far corners, with no seam line, no gradient, no hotspot, no vignette, and no falloff to lighter or darker anywhere in the frame. It is a flat color field, not a photographed backdrop - no surface, no floor, no wall, no corner, no horizon, and no plane the figure stands on or in front of.

Relight from scratch overriding any reference lighting: completely flat shadowless illumination - one enormous soft frontal source at camera position wrapping the subject evenly, matched equal fill from camera-left and camera-right at identical intensity, matched fill from above and below, so both sides of the face read at exactly the same brightness. No key-and-fill ratio, no modelling, no shadow side, no cheek triangle, no nose shadow, no under-chin shadow, no rim light, no hair light, no kicker, no specular hotspot. Extremely low contrast, even, milky, catalogue-flat. Form is described by bone structure, hair strands, and fabric folds alone, not by light and shadow.

Absolutely zero shadow anywhere outside the subject. No cast shadow, no contact shadow, no floor shadow, no drop shadow, no ambient occlusion where the body meets the background, no soft darkening behind the shoulders or beneath the hem, no halo, no edge darkening, and no rim separating the figure from the field. Shading exists only on the subject and stops cleanly at the silhouette. Absolutely zero light bleed outside the subject - no spill, no glow, no bounce, no reflected color cast thrown from the skin or the wardrobe onto the background, and no brightening anywhere behind the figure.

Skin reads matte and velvety - zero shine on forehead, nose bridge, cheekbones, temples, and chin, no oily T-zone. Skin renders at its true natural skin tone and wardrobe at its true natural color, warmth preserved and natural against the neutral gray, never pale or washed-out or cool-shifted by the background. Real peach fuzz at the jaw and hairline, real soft fine even pore texture, subsurface scattering reading as semi-translucent biology, real hair rendered strand by strand with fine flyaways at the hairline, real fabric weave and drape, never plastic, never waxy, never glass-skin, never harsh - fine flattering texture that keeps the face looking good, no acne, no blemishes, no rough pores.

Even sharpness edge to edge across the entire frame. No depth-of-field falloff, no bokeh, no background blur, no lens vignette, no lens distortion, no flare, no bloom, no chromatic aberration, no atmospheric haze, no air between the subject and the background. Photographed on a 50mm prime, soft natural film grain. Photographed not generated.
03Вещи

Карточка костюма

Собери фото всех вещей, что будут на ней (как карточки товара на маркетплейсе): одежда, обувь, ракетка. Фронтально, на чистом фоне, как в карточке товара.

Описывать вещи нужно по геометрии и конструкции — высокий воротник-стойка, рубчик трикотажа, широкий пояс на леггинсах, подошва в ёлочку, форма обода. Нейросеть понимает форму. Логотипы и разметку несут сами референсы, их не надо описывать словами.

Карточка вещей на белом
Карточка вещей на белом

Прикрепить к промпту

  1. Топ спереди
  2. Леггинсы спереди
  3. Кроссовок сбоку
  4. Кроссовки сверху
  5. Ракетка
Промпт карточки вещей — Nano Banana Pro
A clean product flat-lay collage of the items from the attached references, arranged together in a single tall vertical frame on a pure white background.

The pale blue sleeveless top sits at the top of the frame, seen from the front, holding its full worn three-dimensional shape as if on an invisible body, the high mock neck standing open and the cropped hem holding its natural curve. The black leggings sit directly below it, front on, holding their worn leg shape with the high waistband flat and the ankle hems intact, the top's hem and the waistband nearly meeting so the two read as one outfit. The white tennis shoes sit at the lower left, angled three-quarters toward camera, one slightly ahead of the other. The tennis racket lies at the lower right, angled diagonally with the head up and the handle toward the bottom corner, the string bed facing camera. The items are spaced evenly with clean white gaps between them, none overlapping, all fully within the frame.

Keep every item exactly as shown in its reference - the same colour, the same fabric and finish, the same cut, the same seams and panels, the same sole construction, the same frame shape and colour layout, the same string bed, the same markings and graphics in the same positions at the same scale, nothing added, nothing removed, nothing redrawn.

The background is a single flat pure white field - one uniform value at every pixel corner to corner, no seam line, no gradient, no hotspot, no vignette. It is a flat colour field, not a photographed surface - no floor, no wall, no corner, no horizon, and no plane the items rest on.

Completely flat shadowless illumination - one enormous soft frontal source at camera position, matched equal fill from every side, no key-and-fill ratio, no modelling, no specular hotspot. Absolutely zero shadow anywhere outside the items - no cast shadow, no contact shadow, no drop shadow, no ambient occlusion, no halo, no edge darkening. Shading exists only on the items and stops cleanly at their silhouettes.

Real fabric weave and knit texture with visible compression in the leggings, real mesh texture and rubber surface on the shoes, real string tension and frame surface on the racket. Even sharpness edge to edge, no depth-of-field falloff, no bokeh, no lens vignette, no distortion, no flare. Photographed on a 50mm prime, soft natural film grain. Photographed not generated.
04Корт

Плейт локации

Локацию тоже лучше зафиксировать картинкой, а не описывать заново в каждом промпте. Один плейт держит покрытие, разметку, сетку, ограждение, деревья и трибуны — и дальше корт перестаёт плавать между планами.

Референсов тут нет, всё собирается с текста. Свет в плейте задаётся тот же, что будет в ролике: если в видео фиксируется низкое солнце, оно должно быть и здесь.

Плейт грунтового корта
Плейт грунтового корта
Промпт локации — Nano Banana Pro
A wide establishing photograph of an outdoor clay tennis court in late afternoon, in a tall vertical frame.

Deep terracotta clay surface with fine crisp white lines, freshly swept with faint drag-brush arcs across it. A black mesh net strung across the middle on white-capped posts. A tall dark green windscreen fence running the length of the far side, a low weathered white wall with a simple manual scoreboard behind it, and a row of tall cypress and pine beyond, dry and dusty. Empty aluminium bleachers at the frame edge. A pale hazy sky with no clouds low in the frame.

Low late-afternoon sun raking in from one side at a shallow angle, long soft shadows stretching across the clay, warm light catching the top edge of the net cord and the dust hanging in the air. Warm terracotta and olive palette, gentle atmospheric haze in the far trees.

Photographed on a 35mm prime, natural depth, no people anywhere in frame, no logos, no signage text. Soft natural film grain. Photographed not generated.
05Картапо желанию

Карточка персонажа из трёх панелей — по желанию

Для ролика хватает трёх опорных картинок: лок лица из шага 02, карточка вещей из шага 03 и плейт локации из шага 04.

Проблема вылезает, когда генераций становится несколько. Всё, что нигде не показано надетым, нейросеть додумывает заново каждый раз. У меня, к примеру, на двух генерациях носки вышли разной длины, а на одной героиня оказалась с распущенными волосами, хотя в остальных кадрах было иначе.

Карточка персонажа закрывает ровно эти мелочи: длину носков, посадку пояса, где заканчивается топ, украшения, как лежат волосы. Она показывает образ надетым в трёх ракурсах: фронт без головы, спина целиком, лицо крупно. Но, опять же, качественный результат можно получить и без неё.

Прикрепить к промпту

  1. Полноростовая сборка образа
Промпт карты — Nano Banana Pro, 3 панели
A three-panel character reference sheet composed as one horizontal frame, divided into three equal vertical panels side by side, thin clean separation between panels, the same figure and the same outfit rendered identically across all three. No text, no labels, no numbering anywhere in the image.

Lean athletic build, warm golden skin with a soft matte finish. Long near-black hair worn loose and swept back off the face at the crown, straight with a soft natural bend, falling down both sides past the collarbone - the hair is loose in every panel, never tied, never in a ponytail, never in a bun, never braided. Large almond hazel-brown eyes, thick straight dark brows, full lips, bare-skin makeup with natural lash and brow grooming. A cluster of three small dark beauty marks on the left cheek below the outer corner of the eye. Short natural nails, no polish.

She wears a pale blue sleeveless cropped top with a high mock neck, a smooth close-fitting ribbed knit, wide shoulder straps cut in at the armhole, and a hem ending just above the navel with a narrow band of bare midriff below it. Black full-length leggings, matte compression knit, a wide high waistband sitting above the hipbone and flat across the stomach, the hems ending at the ankle bone. White tennis shoes with a ribbed mesh upper, a wide white midsole, a black rubber toe and heel, a black herringbone outsole and white laces. A fine silver chain at the collarbone and exactly two thin silver hoops in the left ear, nothing in the right ear, no other jewelry anywhere.

LEFT PANEL - full body front view, no head, no neck, and no hair. The body stands squared to camera from the shoulders down to the shoes, arms relaxed at the sides, hands open and loose, weight even across both feet. There is no head, no neck, and no hair at all - nothing rises above the shoulder line, and no hair falls across the chest or shoulders. The high mock neck stands up and holds its own three-dimensional shape at the top of the garment, its opening an empty dark hollow looking down into the inside of the top. The outfit reads as worn by an invisible body - full volume, natural drape, real fabric tension across the chest and shoulders, the leggings holding their leg shape, but nothing emerging from the collar. No stump, no skin, no cut edge, no anatomy, not blurred, not faded, no ghosting. The panel keeps full headroom so the figure sits at the same scale and position as a normal full-body portrait.

CENTER PANEL - full body rear, head attached. The complete figure from above the crown down to the shoes, seen directly from behind, standing squared away from camera, weight even, arms relaxed at the sides. The loose hair falls down the back past the shoulder blades, parting to show the shoulder line of the top and the bare strip of back between the hem and the waistband. The high waistband and the seam running down the back of each leg read clearly, and the shoe heels and outsoles are fully visible.

RIGHT PANEL - tight chest-up face lock. Framed from just above the crown down to the collarbones with the face filling the panel, squared to camera, head level, eyes directly to camera, neutral model face-card expression, lips closed and relaxed. The mock neck collar and the silver chain enter at the bottom of frame. Every facial plane, the full eye construction, brow, nose, lip shape, hairline and the three beauty marks read at maximum detail.

Skin renders at its true natural skin tone, identical in value and hue across the face, arms, hands, back, and body in every panel, never darkened, never tanned, never pale or washed-out or cool-shifted.

The background is a single flat 18% neutral gray field across all three panels - one uniform value at every pixel, identical beside each figure and in the far corners, with no seam line, no gradient, no hotspot, no vignette, and no falloff anywhere. It is a flat color field, not a photographed backdrop - no surface, no floor, no wall, no corner, no horizon, and no plane the figures stand on.

Completely flat shadowless illumination, uniform across all three panels - one enormous soft frontal source at camera position wrapping each figure evenly, matched equal fill from camera-left and camera-right at identical intensity, matched fill from above and below. No key-and-fill ratio, no modelling, no shadow side, no nose shadow, no under-chin shadow, no rim light, no hair light, no specular hotspot. Extremely low contrast, even, milky, catalogue-flat.

Absolutely zero shadow anywhere outside the figures in any panel - no cast shadow, no contact shadow, no floor shadow beneath the feet, no drop shadow, no ambient occlusion, no halo, no edge darkening. Shading exists only on the figures and stops cleanly at each silhouette. Absolutely zero light bleed outside the figures - no spill, no glow, no bounce, no reflected color cast onto the background.

Skin reads matte and velvety with real fine even pore texture, peach fuzz at the jaw and hairline, subsurface scattering, hair rendered strand by strand, real ribbed knit texture in the top and matte compression knit in the leggings, real mesh and rubber texture on the shoes. Never plastic, never waxy, never glass-skin. No acne, no blemishes, no rough pores.

Even sharpness edge to edge across the entire frame. No depth-of-field falloff, no bokeh, no background blur, no lens vignette, no distortion, no flare, no bloom, no chromatic aberration. Photographed on a 50mm prime, soft natural film grain. Photographed not generated.
06Видео

Генерация в Seedance

Длинный неупорядоченный промпт даёт случайный результат. В моём же, который даю ниже, работает жёсткий каркас: сначала техническая эстетика, затем блоки CRITICAL с тем, что не должно сломаться никогда, затем прикреплённые материалы с явным указанием, что каждый из них несёт, затем геометрия сцены, замки линз по планам, свет, атмосфера, тайминг по секундам, физика, актёрская задача, звук и финальная цепочка Locks.

Два приёма, которые стоят половины результата. Первый: у каждого прикреплённого тобой материала прописана его единственная роль, и явно сказано, чего он не несёт — иначе видео отдаст свой костюм, а карточка вещей свой свет. Второй: длинные непрерывные розыгрыши (обмены ударами) на 4–5 секунд вместо нарезки коротких кусков. Именно они дают ощущение живой игры, а не набора поз.

Гони по одному плану за раз, а не всю двадцатку сразу — брак дешевле выкидывать поштучно.

Кадр из готового материала
Кадр из готового материала

Прикрепить к промпту

  1. Видео-источник — лицо, волосы, костюм
  2. Плейт локации — корт и окружение
Промпт ролика — Seedance 2.5, 30 секунд
9 shots, 30 seconds: 0.0–2.0, 2.0–6.0, 6.0–8.5, 8.5–13.5, 13.5–16.0, 16.0–19.0, 19.0–24.0, 24.0–26.5, 26.5–30.0. Hard cuts, no transitions, no dissolves. Real-time throughout, no slow motion, no ramping.

Style: 8K photorealism, organic 65mm grain and halation, high dynamic range, large-format film. NOT a 3D render, NOT a game engine, NOT a cartoon. Intimate handheld closeness, matte materials. Skin at pore-level realism — visible pores, vellus hair, sweat sheen, no retouching. 24fps, true 180-degree shutter at 1/48 second, genuine motion blur. No flicker, no warping, no morphing, no interpolation, no ghosting.

NO ON-SCREEN TEXT — CRITICAL: no text of any kind anywhere in frame at any point. No captions, no subtitles, no lower thirds, no titles, no credits, no watermarks, no timecode, no corner markings, no logos, no UI overlays, no Chinese or Korean characters. Any text or graphic in the video source is not carried forward.

SHE IS ALONE — CRITICAL: she is the only person in the sequence. No opponent ever — not across the net, not in the distance, not as a silhouette, not blurred, not partially in frame, not as a shadow, not even in the long wide shots where a rally is running. No coach, no ball kids, no spectators, no figures behind the fence, no bodies on the bleachers, no limbs entering from any edge. Balls arrive from off-frame and leave to off-frame with nobody sending or receiving them.

@Video 1 = the source for the woman and everything she wears — her face, bone structure, skin tone, build, hair, and her full outfit exactly as it sits on her body: every garment, its cut, colour, fit and hem position, her footwear, socks, jewellery and racket. All of it continues unbroken, identical from the first frame to the last, never changing between shots. Nothing is redesigned or substituted. No location, background or graphics are taken from this source.

@Image 1 = environment reference, the sole source of the location — court surface and texture, white lines, net and posts, perimeter fence, low wall, trees beyond, empty bleachers. The whole sequence plays on this court and nowhere else.

GEOMETRY MAP: the court runs left to right, the net at mid-depth, the fence and low wall behind it, the treeline furthest back. She works the near side across the full depth of the court and never crosses to the far side. The bleachers sit at the frame edge, empty.

FIRST FRAME: the racket already turning in her hand, mid-rotation. No empty establishing frame, no static hold.

LENS LOCK SHOT 1 = 12° (200mm) tele detail, tight on the hand and racket.
LENS LOCK SHOT 2 = 34° (65mm) short tele, wide of the court at eye level.
LENS LOCK SHOT 3 = 63° (35mm) reportage wide, low, travelling laterally.
LENS LOCK SHOT 4 = 8° (300mm) extreme long lens from far down the court.
LENS LOCK SHOT 5 = 84° (22mm) classic wide, high above the court looking down.
LENS LOCK SHOT 6 = 84° (22mm) classic wide, lens on the court surface beneath the net.
LENS LOCK SHOT 7 = 47° (50mm) neutral, shooting through the fence mesh.
LENS LOCK SHOT 8 = 63° (35mm) reportage wide, low, travelling toward her.
LENS LOCK SHOT 9 = 29° (80mm) portrait compression, close behind her back.
No focal drift mid-shot.

CAMERA: heavy handheld throughout, 12–25° of swinging cant that never passes through level and never settles square, shoulder-mounted mass, weight shifts, breath, over-correction a beat behind the body. In the long rallies the operator pans and re-frames constantly to keep her in frame, losing her on the fastest changes of direction and catching her again. Every frame mid-move but the eye can still land. Never locked, never stabilized, never gimbal-glide, never drone.

LIGHT: high hard sun holding the same direction for all 30 seconds, bright specular hits on the shoulders and crown, short compact shadows pooling beneath the body, strong contrast on the court, clean white bloom in the sky. The sun never drops to the horizon, never becomes warm evening light, never overcast, never night.

ATMOSPHERE: clean bright air with real density at distance — foreground razor sharp on skin, fabric and fence mesh, her body crisp in the mid-ground, the far fence and trees softening into haze. Dust lifts only where a shoe scuffs the surface or a ball skids off it, and settles immediately. No plumes, no fog-machine texture, no volumetric shafts, no god rays.

0.0–2.0s (SHOT 1, the spin): tight tele detail on her hand at hip height, the racket spinning by the grip through two full rotations, the frame flashing sunlight as it turns, her fingers catching it and closing hard around the handle.
2.0s HARD CUT
2.0–6.0s (SHOT 2, the baseline rally): a single unbroken wide of the near court at eye level, the camera panning hard with her. She plays a continuous rally alone — a ball flies in from off-frame to her forehand and she drives it back out of frame, splits, pushes hard left as the next arrives and digs it out low on the backhand, recovers to the middle with quick crossover steps, then takes a third high on the rise and punches it away. Three clean strikes with no cut, the body never still, hem and hair snapping through every change of direction, grit spraying off each plant.
6.0s HARD CUT
6.0–8.5s (SHOT 3, the low run): the camera low to the ground travelling laterally alongside her at speed as she sprints along the baseline, the surface streaking past in the foreground, only the lower half of her body in frame — legs driving, shoes punching off the court, grit kicking back toward the lens, the fence smearing into motion blur behind.
8.5s HARD CUT
8.5–13.5s (SHOT 4, the distance): a single unbroken extreme long lens from far down the court, she plays small and compressed deep in the flattened frame. A second continuous rally — sprinting wide to reach a ball and whipping it back, recovering across the baseline at speed, sliding into a stop and driving another, then chasing a short ball forward and lifting it away. Heat shimmer between the lens and her, the empty far court and empty bleachers filling the frame around her.
13.5s HARD CUT
13.5–16.0s (SHOT 5, from above): high above the court looking down, she covers the width of the near side in a fast diagonal, sliding into a stop and hitting on the stretch, her compact shadow tracking with her across the lines, the whole empty court laid out around her.
16.0s HARD CUT
16.0–19.0s (SHOT 6, under the net): the lens resting directly on the court surface beneath the net cord, the ground filling the bottom of frame in razor focus, the black net band cutting across the top. Her legs work in the mid-ground beyond it, shoes hammering the surface, a ball skidding low past the lens and away. She drives forward at the net, her shoes filling the frame as she plants and pushes off, grit flicking straight at the lens.
19.0s HARD CUT
19.0–24.0s (SHOT 7, through the fence): a single unbroken take shot through the dark perimeter fence mesh, the mesh sharp and geometric in the foreground, her body reading clearly behind it. A third continuous rally — she hits a forehand hard, sprints right behind the mesh as the pattern slides across her, digs out a low backhand, recovers with a crossover and drives another, then breaks forward away from the fence. Four strikes with no cut, her shape moving in and out of the mesh lines throughout.
24.0s HARD CUT
24.0–26.5s (SHOT 8, straight at the lens): the camera low on the ground travelling toward her as she sprints directly at it, the court rushing past beneath the frame, her legs scissoring and shoes slamming down closer and closer until she plants hard beside the lens and swings, the frame shuddering with the impact.
26.5s HARD CUT
26.5–30.0s (SHOT 9, behind the back): close behind her shoulders, camera riding her through the last exchange — she steps in and rips a two-handed backhand, hair whipping across her back, recovers with a hard crossover step and drives a forehand away, chases one more ball forward and stabs it past the net, then holds her finish with her chest heaving and her back to camera, the empty court stretching out in front of her.

PHYSICS: real bodyweight around 60kg. Shoes bite and skid on the court, knees absorbing every landing, weight shifting through the hips on recovery. The ball compresses into the string bed and the strings rebound, the frame twisting in her grip on off-centre hits. Hair and hem lag the motion and settle a beat behind it. Accurate contact shadows where the shoes meet the surface. Sweat beads at the hairline and runs. Nothing floats, nothing slides.

ACTING: effort held in the body — jaw set, nostrils flaring on the breath, brow drawn down through the strikes, eyes tracking each incoming ball and locking to a target ahead of every stroke, lips parted and breathing hard, fatigue building across the rallies. No eyeline to camera at any point. No teeth-showing smile.

AUDIO: fully diegetic. The squeak and skid of shoes on the court, the flat crack of the ball off the string bed on every strike, the ball skidding on the surface, the whirr of the racket spinning in her hand, the rattle of the fence, hard breathing and short grunts of effort, wind in the trees. No music of any kind — no score, no soundtrack, no underscore, no swell, no drone, no pad, no strings, no percussion, no beat. No dialogue, no voiceover, no crowd noise, no applause.

LOCKS: the video holds her face, hair and entire outfit with its exact fit in every frame, unchanged between shots; the environment reference holds the court, lines, net, fence, wall, trees and empty bleachers; the high hard sun holds its direction and short shadows for all 30 seconds; the far court stays empty in every shot including all three long rallies; no text carries forward; the handheld register holds throughout; the lens lock holds per shot; real-time speed holds in every shot, all three long rallies playing continuously with no internal cut.

Что ломается чаще всего

Четыре вещи съедают больше всего генераций. Если держать их под контролем, весь процесс проходит почти без брака.