СПЕШУ АИ: Проводники в ИИ | ChatGPT, Claude, Gemini и ещё 300+ нейронок
3.58K subscribers
331 photos
196 videos
3 files
251 links
SpeShu.AI — платформа с мультидоступом к 300+ нейросетям

Открыть сайт 👉 https://speshu.ai
Download Telegram
Промпт для профессионального рекламного видео в Seedance 2.0:

01 [0:00–0:02.5] SCENE: VOID | ACTION: 50mm wide, pure matte black empty frame, the headphones fade into focus suspended in dead-centre, no light source visible, only the rim-light catching the chrome ring around the papaya disk | SOUND: deep sub-bass pulse, total silence between hits
02 [0:02.5–0:05.0] SCENE: FIRST RIPPLE | ACTION: 50mm wide, a single concentric chrome ripple emanates outward from the earcup like a sonar wave, papaya disk pulses in sync | SOUND: low sub-bass thump, reverse-reverb sweep
03 [0:05.0–0:07.5] SCENE: HARMONIC | ACTION: 100mm macro on the disk, three nested chrome ripples passing through frame at different frequencies, papaya disk glowing warmer with each pulse | SOUND: tubular bell strike, harmonic overtones decay
04 [0:07.5–0:10.0] SCENE: BASS DROP HERO | ACTION: 100mm macro extreme slow-mo, the moment a single low-frequency chrome ripple snaps outward from the earcup at high amplitude — hero beat — papaya disk flares like a sun in eclipse, chrome ring pulses outward | SOUND: bass-drop hit, tubular bell sustains, floor-shaking sub
05 [0:10.0–0:12.5] SCENE: ECHO LATTICE | ACTION: 50mm low angle, the headphones suspended above an infinite black floor, multiple chrome ripples creating a lattice pattern emanating outward in all directions | SOUND: bass groove resolves, distant echo tail
06 [0:12.5–0:15.0] SCENE: END CARD | ACTION: clean matte-black plate, headphones suspended hero side-profile, a single warm papaya pulse glowing from the disk, chrome rim-light only | SOUND: final sub-bass tail, total silence

End-card text (panel 06 only): "BOWERS & WILKINS — PX8. HEAR EVERYTHING. [ВАША ССЫЛКА]"


В 200 раз дешевле создать такое видео в SpeShu.AI. Копируйте промпты, ставьте свои продукты и получайте ролики, которые принесут рекордные продажи.
Ошибка №4. Примеры сильнее описаний, но их никто не даёт

Что происходит
Человек объясняет модели словами, какой стиль, какой формат, какой тон. Модель понимает лишь приблизительно.

Что говорят исследования
Few-shot prompting — это техника, в которой модели дают несколько коротких примеров в формате «вход → выход», чтобы она поняла не только задачу, но и нужный способ ответа.

Если zero-shot оставляет модели простор для догадок, то 2–5 хороших примеров часто буквально фиксируют траекторию ответа: как именно классифицировать, какой тон выдерживать, какие поля заполнять, что считать ошибкой.

❌ Плохой промпт: «Напиши 5 заголовков для статьи про ИИ в логистике. Пусть будут цепляющие, но не кликбейтные».
✅ Хороший промпт:
«Напиши 5 заголовков для статьи про ИИ в логистике. Стиль — как в этих трёх примерах с моего сайта:
"Мы внедрили ИИ в сортировку 6 месяцев назад. Вот что реально изменилось"
"Почему автоматизация склада окупается не за год, как пишут, а за три"
"Что говорят водители о диспетчер-боте: три претензии и одна похвала" Общие признаки: конкретика, честность, отсутствие слов "революция", "прорыв", "инновация". Первое лицо множественного числа или прямая цитата. Длина — 8-14 слов».
Ошибка №5. «Не пиши...» — плохая формулировка (но с важной оговоркой)

Что происходит
Пользователь просит: «Только не пиши шаблонно», «не используй слово "уникальный"», «не делай списков». Результат часто игнорирует запрет.

Что говорят исследования
Для текстовых LLM (ChatGPT, Claude) негативные инструкции работают частично, доказывают составители Prompt Engineering Guide. «Пиши короткими абзацами» работает лучше, чем «не пиши длинные абзацы».

Но это не абсолют. В техническом контексте негативные инструкции работают: «Не включай персональные данные», «Не выдумывай источники, если не уверен» — модель этому следует.

❌ Плохой промпт: «Напиши статью. Не делай её скучной, не используй канцелярит, не лей воду, без списков, не делай длинных предложений».
✅ Хороший промпт: «Напиши статью в разговорном стиле, с короткими предложениями (средняя длина — 10-12 слов). Используй конкретные примеры и цифры. Формат — сплошной текст абзацами, без буллитов. Целевой объём — 600 слов».
This media is not supported in your browser
VIEW IN TELEGRAM
Pixar умер. Вот два промпта, которые делают мультики на 30 МЛН просмотров

Чтобы получать топовый результат, прикладывая минимум усилий, используйте нашу двухчастную структуру.

Сначала сгенерируйте раскадровку (фото) и только потом напишите промпт в видеогенератор. Обязательно приложите раскадровку при этом.

Как сделать подобную раскадровку в ChatGPT Image 2.0?

Create a professional animation storyboard sheet in a hand-drawn cartoon style, using the attached reference only as layout inspiration.

Show a sequence where a chef cooks a dish and then gives the finished plate to another character.

The final image must look like a real storyboard presentation page for an animated film.

Layout: a full storyboard sheet on a white page, with multiple small rectangular storyboard panels arranged in a grid. Each panel shows a different step of the sequence.

Sequence of actions:

chef prepares ingredients
chef cuts or arranges ingredients
chef starts cooking in a pan
close-up of sizzling food
chef plating the dish
chef checks the final result
chef walks toward another character with the plate
chef presents the dish
second character receives the plate
reaction shot of the second character

Visual style: hand-drawn storyboard sketches, cartoon animation style, expressive characters, loose pencil/ink lines, rough but readable drawings, like a real pre-production storyboard.

Important storyboard details:

each panel has a shot number
small handwritten production notes near the frames
arrows showing motion and camera movement
red frame borders around storyboard panels
blue arrows and direction markings
notes section at the bottom of the page
character design/reference section on the right side of the sheet
small character turnaround sketches or portrait references on the right
optional color palette swatches in a corner

Composition: the image should look like one complete storyboard page, not separate images.

Mood: charming, cinematic, animated-feature preproduction style.

Keep it clean, readable, and visually similar to a professional animation storyboard board.
Как оживить в Seedance 2.0?
Create a short animated sequence in a charming cartoon style, based on a professional storyboard-like scene.
Main сюжет: a chef prepares a dish, cooks it, plates it, then walks over and gives the finished plate to another character.
Style: animated film look, expressive cartoon characters, warm kitchen atmosphere, smooth motion, clear storytelling, slightly stylized but readable.
Maintain character consistency across all shots.
0s – 1.5s
Wide establishing shot of the kitchen.
The chef stands at the counter with ingredients laid out.
Warm lighting, cozy animated kitchen, slight ambient movement.
1.5s – 3s
Medium shot.
The chef quickly chops and arranges ingredients on the table.
Energetic hand motion, lively cartoon timing.
3s – 4.5s
Close-up shot.
The chef drops ingredients into a hot pan.
Food sizzles, steam rises, dynamic cooking motion.
4.5s – 6s
Close-up of the pan.
The dish is being stirred and cooked.
Add steam, little sparks of heat, appetizing motion.
6s – 7.5s
Medium shot.
The chef plates the finished dish carefully and красиво.
Focus on the final presentation.
7.5s – 9s
Medium close-up.
The chef looks proudly at the finished plate, smiles slightly, then turns.
9s – 11s
Wider shot.
The chef walks toward the second character, holding the plate with both hands.
The second character waits nearby, watching with interest.
11s – 13s
Medium shot on both characters.
The chef offers the dish to the second character.
The second character receives the plate.
13s – 15s
Reaction shot.
The second character looks pleased and impressed.
The chef stands beside them, proud and satisfied.
End on a warm, charming animated moment.
Camera language: cinematic storyboard-style cuts, clear visual storytelling, wide / medium / close-up shot variation.
Motion: smooth animation, expressive gestures, readable actions, no chaotic movement.
Mood: warm, friendly, playful, animated-feature style.


Обе нейросети можно протестировать в SpeShu.AI без зарубежных подписок (по 20$ каждая).

🎁 А бонусом вы получите 15% к сумме вашего пополнения, если введёте секретный промокод нашего канала: TG15
Сфотографируйтесь на обложку журнала в стиле Y2K

Такие фото набирают по 100 000 просмотров за сутки.

Инструкция:
1. Зарегистрируйтесь на speshu.ai
2. Введите промокод TG15 в личном кабинете и получите 15% к пополнению 🎁
3. Откройте ChatGPT Image 2.0, прикрепите своё фото и вставьте промпт:

High-fashion Y2K-inspired editorial artwork featuring a stylish young East Asian woman standing confidently at the center of the composition. She has luminous dewy skin, subtle glam makeup, and flowing dark hair with airy see-through bangs. Outfit styled with modern Korean street-fashion aesthetics and layered trendy accessories. The background is packed with bold experimental typography, colorful magazine-inspired graphics, retro pop visuals, chrome accents, playful icons, stickers, hearts, stars, speech bubbles, UI overlays, and barcode elements.
Design feels like a nostalgic early-2000s fashion magazine cover blended with contemporary K-style campaign visuals. Include scattered Japanese and Korean-inspired text details, dynamic layered compositions, vibrant commercial poster energy, crisp studio flash lighting, strong contrast, editorial photography realism fused with graphic design aesthetics, highly detailed textures, razor-sharp focus, luxury print-quality finish, visually immersive, ultra-premium fashion campaign look, cinematic composition, 8K resolution.
This media is not supported in your browser
VIEW IN TELEGRAM
Соцсети разделились на два лагеря из-за нового ИИ-тренда

Одна часть людей генерирует гламурные видео со стадиона со своим лицом, а другая категорически против. По их мнению, участники тренда теряют свою индивидуальность.

Мы нашли компромисс. О нём — в следующем посте. А пока покажем, как повторить тренд.

Загрузите фото в ChatGPT Image 2.0 на сайте SpeShu.AI (не забудьте о бонусе — 15% к пополнению по промокоду TG15). Затем вставьте промпт:

Сохранить лицо загруженного человека в исходном виде. Избегать стиля «AI-красавца/красавицы». Создать изображение, которое выглядит как «случайный кадр с обычным зрителем во время прямой трансляции матча KBO (Корейской бейсбольной лиги)».
Сеттинг и композиция:
Стиль скриншота трансляции канала SPOTV KBO.
Ракурс «фанкама» или трибунной камеры.
Естественное окружение: другие зрители на фоне, стаканы с пивом, атрибутика болельщиков, ручные вентиляторы.
Важные ограничения (ЧТО ЗАПРЕЩЕНО):
Никакой чрезмерной ретуши лица.
Запрещено увеличивать глаза.
Запрещено корректировать челюсть/подбородок.
Никакого эффекта «фарфоровой» или размытой кожи.
Избегать стиля модной фотосессии или образа «инфлюенсера».
Эффект реальной трансляции:
Качество живого эфира: легкая расфокусировка, артефакты сжатия, минимальный смаз (motion blur).
Реалистичная текстура кожи: поры, мелкие волоски, жирный блеск от пота.
Естественная поза: человек сидит расслабленно, закинув ногу на ногу, смотрит матч.
Взгляд: естественный, будто человек едва заметил камеру или не смотрит в неё вовсе.
Ключевая цель: Это не должен быть «человек, созданный нейросетью». Это должен быть «обычный человек, который случайно попал в кадр и стал виральным в соцсетях».


Чтобы сделать видео, прикрепите сгенерированное фото в Kling 3.0 на SpeShu.AI и промпт:

Add gentle, believable motion to the main subject in the center of the frame, making it feel like the person is being shown live by a real stadium broadcast camera during a busy baseball game. The subject should respond to the action naturally, with realistic eye direction, small head movements, blinking, subtle breathing, slight shifts in expression, and spontaneous reactions typical of someone watching the game from the stands.
The behavior should feel candid and unposed, as if the broadcast camera briefly picked out this person in the crowd during a live telecast. The audience around them should also appear active and convincing, with natural crowd movement, an excited game-day atmosphere, and the visual feel of an authentic sports broadcast.
Keep the overall result immersive, cinematic, and highly realistic — like a genuine moment from a televised baseball game. Preserve a natural posture, subtle blinking, and true live sports broadcast energy. Avoid making the eyes look crossed.


Какой компромисс мы нашли, расскажем в следующем посте в 15:00. Спойлер: вы наберёте охваты с помощью тренда, но не потеряете свою уникальность ❤️
This media is not supported in your browser
VIEW IN TELEGRAM
Подумайте, что ещё может происходить на стадионе, — тренд заиграет новыми красками

Вот такой забавный промпт, чтобы не просто сидеть на трибуне, а включиться в игру на 1 000%:

Prompt: Ultra-realistic sports broadcast still of a glamorous woman sitting in a packed football stadium crowd during a night match, wearing a dark brown sleeveless high-neck satin top and black square earrings, shoulder-length light brown/blonde hair styled in soft waves. She is casually drinking from a tall blue aluminum can while holding a half-eaten cheeseburger in the other hand. Around her are fans in bright yellow and blue football jerseys and scarves, creating strong team-color contrast. The scene feels candid and cinematic, captured mid-game from a TV broadcast camera angle with shallow depth of field. Include realistic stadium seating, crowded audience atmosphere, broadcast overlay graphics in the top-left corner showing a live football score and match timer, and a sports network watermark in the top-right. Natural arena lighting, detailed skin texture, sharp focus on the woman, slightly blurred background crowd, authentic live sports broadcast aesthetic, 16:9 composition.
This media is not supported in your browser
VIEW IN TELEGRAM
Станьте персонажем игры. Видео на 2 МЛН просмотров

Мы нашли, как повторить в России и стать родоначальником тренда.

Запустите ChatGPT Image 2.0 в SpeShu.AI без зарубежной подписки, вставьте своё фото и промпт:

Сгенерируй интерфейс RPG-игры.
⦁ Модель: Основана на внешности пользователя (добавьте описание черт лица: цвет глаз, волос, форму носа и т.д.).
⦁ Телосложение: Мускулистое, покрыто множеством детализированных черно-белых татуировок (абстрактные узоры, геометрические фигуры, надписи).
⦁ Руки:
⦁ Левая: Кибернетический протез с видимой внутренней проводкой, суставами, поршнями и датчиками. Окрашен в темно-серый металл с акцентами из синего и зеленого света.
⦁ Правая: Более массивная кибернетическая рука с четырьмя длинными, острыми когтями-лезвиями на конце. Покрыта защитными пластинами и каналами для энергии.
⦁ Одежда: Черные шорты, белые носки, кремовые кроссовки.
⦁ Аксессуары: Серебряная цепочка на шее, кольца в ушах.
⦁ Поза: Сидит в позе лотоса на белой подставке. Справа от него стоит красная банка с напитком, слева - разбросанные окурки.
⦁ Стиль: Гиперреализм, студийное освещение.
Затем оживите это фото в Seedance 2.0. Чтобы сэкономить, примените в личном кабинете промокод TG15:

Animate the uploaded image as a realistic live computer interface. Keep the original composition, layout, colors, typography, character design, background, and all UI elements from the source image. The image should look like an actual screen recording or a real interactive application, not a fantasy scene.

Duration: 10 seconds.
Style: realistic interface animation, subtle screen glow, smooth cursor movement, natural micro-interactions, clean UI motion, high-quality digital animation.
Camera: static camera, no zoom, no rotation, no perspective change. The whole frame remains stable as if recorded directly from a screen.
Preserve identity: the character must remain visually consistent with the original image. Do not change face structure, hairstyle, clothing, colors, UI design, text placement, or background.

Storyboard:

0.0–1.0 sec:
The image is static for a brief moment, like a paused interface. A white mouse cursor appears near the lower-right or side area of the screen. Very subtle screen glow and tiny UI shimmer make the interface feel alive. The character is idle, blinking naturally once, with a neutral expression.

1.0–2.0 sec:
The mouse cursor moves smoothly across the interface toward the main interactive area. As it passes over buttons, icons, cards, or panels, they react with subtle hover animations: slight brightening, soft glow, small scale increase, or gentle highlight. The character’s eyes begin to notice the movement of the cursor.

2.0–3.0 sec:
The cursor stops on one UI element, as if hovering over a button or option. The hovered element becomes more highlighted. The character turns their gaze toward the cursor, raises eyebrows slightly, and shows curiosity. The character remains in the same pose and style as the original image.

3.0–4.0 sec:
The cursor clicks once. Show a realistic click animation: small ripple, button press, quick glow, or a tiny UI response. The interface reacts as if selected, but the overall layout must remain the same. The character reacts with a small surprised expression, blinking and slightly opening their mouth.

4.0–5.5 sec:
The character begins speaking naturally, with subtle lip sync and facial expression. The character looks toward the cursor or viewer and says in Russian: “О, ты здесь? Давай посмотрим, что у нас внутри.”
Mouth movement should match speech, with natural blinking and small head movement. Do not exaggerate gestures.

5.5–6.5 sec:
The mouse cursor moves again, slowly and deliberately, exploring another part of the interface. Nearby UI elements respond with hover states: soft glow, thin outline, gentle color change, micro-bounce. The character follows the cursor with their eyes, looking engaged and alive.

6.5–7.5 sec:
The cursor performs a second click on another visible interface element. The selected area briefly highlights. Add subtle animated details inside the UI, such as a small loading dot, soft panel transition, or activated state, while preserving the original photo’s layout and design. No new major windows should appear unless they already exist in the image.

7.5–8.8 sec:
The character smiles slightly and speaks again in Russian: “Видишь? Всё работает как настоящий интерфейс.”
The voice should feel calm, friendly, and conversational. Lip movement is synchronized, expression is warm, eyes occasionally glance at the cursor.

8.8–10.0 sec:
The cursor moves to a final position near the character or main button and stops. The interface remains gently animated with subtle highlights. The character gives a small nod, blinks, and returns to a relaxed idle pose. End on a clean, stable frame that still matches the original image.

Important animation requirements:
- Keep the original photo recognizable and consistent.
- The mouse cursor must look like a normal desktop arrow cursor.
- UI reactions must be subtle and realistic.
- Character animation should be natural: blinking, eye tracking, slight head movement, lip sync.
- No dramatic camera movement.
Продолжение промпта для Seedance 2.0:

- No change to the original layout.
- No extra characters.
- No distorted hands, face, text, or UI.
- Do not invent a completely new interface.
- Do not replace the original background.
- Do not make the animation cartoonish unless the source image is cartoon-style.
This media is not supported in your browser
VIEW IN TELEGRAM
Новая часть пИИксар-анимации — 2.2 МЛН просмотров за сутки

Опять работаем в связке сториборд + раскадровка.

Создаём сторибор в ChatGPT Image 2.0 на сайте SpeShu.AI за 10₽ вместо 1800 рублей (столько стоит подписка):

Create a single horizontal storyboard sheet on a clean white paper background, matching the style and structure of a professional animation storyboard.

The sheet title at the top left:
“DINNER SERVICE CHAOS – SHORT STORYBOARD”

Overall layout:
A wide 16:9 image showing one complete storyboard page.
Use a clean white background with light pencil texture.
Draw everything in rough but readable graphite pencil storyboard style, with subtle blue and red animation planning marks.
Use thin sketchy pencil lines, handwritten labels, small arrows, timing notes, and camera framing boxes.
The page should feel like a real production storyboard pinned on a studio wall.

Main storyboard area:
Place 17 storyboard panels arranged in a grid on the left and center of the page.
Each panel should have a thin rectangular frame.
Each panel should be numbered clearly in small circled numbers from 01 to 17.
Add a short handwritten caption under each panel.
Use red vertical framing lines and small blue arrows to indicate camera movement, breath, energy, or ingredient motion.
The storyboard drawings should be simple black-and-white pencil sketches, not fully rendered illustrations.

Panel layout:
Rows of small storyboard frames across the page, similar to a professional animation storyboard.
Use approximately:
Row 1: panels 01–05
Row 2: panels 06–10
Row 3: panels 11–15
Row 4: panels 16–17, with panel 17 wider than the others as a final wide comedic ending shot.

Right side column:
Create a vertical notes and character design area on the right side of the sheet.
At the top of this column write:
“CHARACTER DESIGN + CASTING”

Add two small character portrait boxes:
1. “THE LADY CHEF”
A small colored character portrait or lightly shaded sketch of an elderly female chef, age 70+, fluffy gray-blue hair, large round glasses, warm eyes, energetic, playful, wise.
Next to her portrait, add handwritten notes:
“AGE: 70+”
“ENERGETIC”
“PLAYFUL”
“WISE”
“MASTER OF FOOD + FLOW”

2. “THE CHEF”
A small colored character portrait or lightly shaded sketch of a male chef, age 30–40, tall chef hat, long face, thin mustache, worried eyebrows, proud but easily unnerved.
Next to his portrait, add handwritten notes:
“AGE: 30–40”
“PROUD”
“COMPETITIVE”
“EASILY UNNERVED”

Below the character boxes, add a small notes section titled:
“NOTES / THEMES”
With handwritten bullet points:
“BREATH = POWER”
“FOOD = ENERGY”
“FLOW = CONTROL”
“EGO = IMBALANCE”

Also add a tiny legend near the top:
A small red square labeled “CAMERA / FRAMING”
A small blue arrow labeled “MOVEMENT / BREATH / ENERGY”

Storyboard panel content:

Panel 01:
Wide establishing shot of a fancy restaurant. Diners sit at elegant tables. The elderly lady chef serves a dish at the center. Caption: “WIDE ESTABLISHING SHOT”

Panel 02:
Close-up of a perfectly plated dish on a white plate. Caption: “CLOSE ON THE DISH”

Panel 03:
Close-up of the elderly lady chef approving the dish through her large round glasses. Caption: “CLOSE ON LADY”

Panel 04:
Close-up of the younger male chef looking smug and confident. Caption: “CLOSE ON CHEF (SMUG)”

Panel 05:
Close-up of the male chef looking worried, something catches his eye. Caption: “CLOSE ON CHEF (WORRIED)”

Panel 06:
Close-up of the elderly lady chef teasing him with a playful mischievous expression. Caption: “CLOSE ON LADY (TEASING)”
Продолжение промпта для ChatGPT Image:

Panel 07:
Wide shot of a busy professional kitchen. The elderly lady chef steps into the kitchen like a conductor entering a stage. Caption: “KITCHEN – WIDE”

Panel 08:
Medium close-up of the lady chef taking a powerful breath, with blue arrows showing airflow and energy around her. Caption: “POWER BREATH”

Panel 09:
Ingredients swirling around the lady chef in controlled motion: tomatoes, herbs, fish, vegetables, utensils, sauce droplets. Caption: “INGREDIENTS IN MOTION”

Panel 10:
Close-up of fresh vibrant ingredients moving through frame in precise arcs. Caption: “CLOSE ON INGREDIENTS”

Panel 11:
Intense cooking scene in the kitchen: flames, pan, steam, chopping, sauce, rapid controlled chaos. Caption: “COOKING – INTENSE”

Panel 12:
Top-down plating shot. Hands arrange food elements perfectly on a plate. Caption: “PLATING – TOP SHOT”

Panel 13:
Return to the restaurant table. The lady chef presents the finished dish while diners and the male chef watch. Caption: “RETURN TO TABLE”

Panel 14:
Close-up of the male chef impressed, giving a reluctant thumbs-up. Caption: “CHEF IMPRESSED”

Panel 15:
Close-up of the elderly lady chef delighted and victorious, smiling broadly. Caption: “LADY DELIGHTED”

Panel 16:
Close-up of the male chef defeated, shoulders slumped, ego punctured. Caption: “CHEF DEFEATED”

Panel 17:
Wide comedic ending shot. Restaurant chaos continues around the table; the lady chef stands proud and happy, the male chef stumbles or slides back in defeat, guests react. Caption: “WIDE COMEDIC ENDING”

Drawing style:
Rough pencil storyboard sketch, loose but confident linework, production-art style, small handwritten annotations.
The drawings must look like previsualization frames, not finished comic panels.
Use mostly grayscale pencil with very subtle blue arrows and red frame marks.
Keep the image clean, readable, and organized.
The right-side character portraits may have light color accents, but the main storyboard panels should remain sketchy black-and-white.

Important:
Do not create a polished comic page.
Do not make full-color final illustrations inside the storyboard panels.
Do not use speech bubbles.
Do not add dialogue.
Do not add extra story panels.
Do not change the order of panels.
Make the page look like a professional storyboard sheet for an animated short film.
Промпт, чтобы анимировать сториборд в Seedance 2.0:

Create a 17-second animated short film based exactly on the provided storyboard image “Dinner Service Chaos – Short Storyboard”.

Follow the storyboard shot by shot, one panel per second. Preserve the same story order, framing, character designs, emotional beats, and comedic rhythm.

Main characters:
1. The Lady Chef, protagonist:
An elderly female chef, age 70+, energetic, playful, wise, expressive, with fluffy gray-blue hair, large round glasses, warm eyes, small stature, confident posture. She is a master of food and flow. She controls chaos through breath, focus, and precise movement.

2. The Male Chef, antagonist:
A male chef, age 30–40, proud, competitive, easily unnerved, with a long face, thin mustache, worried eyebrows, tall chef hat, white chef uniform. He starts smug and confident, then becomes worried, impressed, and finally defeated.

Setting:
A fancy restaurant during chaotic dinner service, connected to a busy professional kitchen. Elegant tables, diners, waiters, warm lighting, refined atmosphere, but with comic chaos underneath. The kitchen is full of counters, pans, plates, vegetables, herbs, sauces, steam, staff, and fast movement.

Visual style:
Polished cinematic animated short, expressive character animation, warm restaurant lighting, lively kitchen motion, clean readable compositions. The film should feel like the storyboard brought to life. Controlled chaos, comedic timing, no dialogue, no subtitles.

Camera:
Follow the exact storyboard framing: wide establishing shots, close-ups, kitchen wide, ingredient close-ups, top-down plating shot, and final wide comedic ending. No random camera movement. No extreme zooms. Each shot must be clear and readable.

Duration: 17 seconds.
Aspect ratio: 16:9.
Pacing: 17 shots, one shot per second.

SECOND-BY-SECOND STORYBOARD:

0–1 sec:
Shot 01. Wide establishing shot of a fancy restaurant. Diners sit at elegant tables. Waiters and kitchen staff move in the background. The elderly lady chef stands near a table and serves a dish with calm confidence. The room feels upscale, busy, and slightly chaotic.

1–2 sec:
Shot 02. Close-up on the dish. A perfectly plated meal sits on a clean white plate. The dish is balanced, elegant, and precise. Soft steam and glossy highlights make it look fresh and delicious. The camera is steady and intimate.

2–3 sec:
Shot 03. Close-up on the elderly lady chef. She examines the dish through her large round glasses. She gives a small approving smile and a tiny satisfied nod. Her expression is wise, playful, and quietly proud.

3–4 sec:
Shot 04. Close-up on the male chef looking smug. He faces camera with a self-satisfied grin, raised eyebrows, and proud posture. He looks like he believes he is in control of the dinner service.

4–5 sec:
Shot 05. Close-up on the male chef becoming worried. His smug smile fades. Something catches his eye off-screen. His eyebrows lift and pinch together. His eyes widen slightly. Add a subtle nervous twitch or swallow. He realizes the situation is changing.

5–6 sec:
Shot 06. Close-up on the elderly lady chef teasing him. She looks mischievous and amused, as if she knows exactly what he is thinking. She gives a playful smile, slight wink or eyebrow raise, with a confident sparkle in her glasses.

6–7 sec:
Shot 07. Wide shot of the professional kitchen. The elderly lady chef steps into the kitchen like a conductor entering a stage. Counters, pans, shelves, hanging utensils, plates, vegetables, cooks, and steam surround her. The kitchen is busy, but she stands centered and calm.

7–8 sec:
Shot 08. Power breath. Medium close-up of the lady chef. She inhales deeply with full focus. Her shoulders rise slightly. Her gray-blue hair and chef clothing subtly lift from the energy. Blue motion lines or soft air-flow trails show breath and power building around her. She becomes intensely focused.

8–9 sec:
Shot 09. Ingredients in motion.
Tomatoes, herbs, fish, meat, vegetables, spices, utensils, sauce droplets, and plates swirl around her in controlled circular motion. She guides them with calm hand gestures and breath. The movement is energetic but precise, like an orchestra.

9–10 sec:
Shot 10. Close-up on ingredients. Fresh vibrant ingredients fly through the frame in elegant arcs. Show tomatoes, herbs, carrots, fish, meat, onions, and sauce drops. Every object looks fresh, colorful, and sharply timed. The motion is fast but readable.

10–11 sec:
Shot 11. Intense cooking. The elderly lady chef cooks rapidly in the kitchen. Flames rise from a pan, sauce splashes in a clean arc, knives chop rhythmically, steam bursts upward, pans flip ingredients. She remains calm and fully in control while the kitchen moves around her.

11–12 sec:
Shot 12. Top-down plating shot. A clean white plate seen from above. Hands place each element perfectly. Garnishes land exactly in place. Sauce draws a beautiful curve. Vegetables, protein, herbs, and final garnish align into an elegant restaurant-quality composition. Every element clicks into position.

12–13 sec:
Shot 13. Return to the restaurant table. The elderly lady chef presents the completed dish to diners. The male chef is nearby, watching in disbelief. Guests lean forward, impressed. The composition mirrors the opening restaurant shot, but now the lady chef clearly dominates the moment.

13–14 sec:
Shot 14. Close-up on the male chef impressed. He stares at the dish and the lady chef with awe. His pride collapses into admiration. He gives a reluctant thumbs-up or small approving gesture. His expression is humbled, surprised, and impressed.

14–15 sec:
Shot 15. Close-up on the elderly lady chef delighted. She beams with joy and victory. Her eyes shine behind her big glasses. She laughs silently or opens her mouth in a delighted smile. Her fluffy hair and glasses bounce slightly with her happiness.

15–16 sec:
Shot 16. Close-up on the male chef defeated. His shoulders slump. His face droops. His mustache and eyebrows emphasize defeat. His tall chef hat may sag slightly. He looks humbled, embarrassed, and drained, as if his ego has been punctured.

16–17 sec:
Shot 17. Final wide comedic ending in the restaurant. The elderly lady chef stands proudly near the table, delighted by the chaos. The dining room is active with comic movement: guests reacting, waiters rushing, background staff moving quickly, the male chef stumbling or sliding backward in defeat. Despite the chaos, the lady chef and the dish remain perfectly composed. End on a triumphant funny tableau.

Core theme:
Breath equals power. Food equals energy. Flow equals control. Ego equals imbalance.

Important requirements:
- Follow the provided storyboard exactly.
- Keep the 17-shot order.
- One shot per second.
- Preserve the elderly lady chef and male chef designs.
- Preserve the restaurant and kitchen setting.
- No dialogue.
- No subtitles.
- No random text.
- No extra main characters.
- Keep the tone comedic, energetic, elegant, and readable.
- The lady chef must always feel in control.
- The male chef must clearly transition from smug to worried to impressed to defeated.

Negative prompt

Do not change the shot order, do not skip storyboard panels, do not add new major scenes, do not add dialogue, do not add subtitles, do not add random text, do not turn the scene into a horror film, do not make the tone serious or tragic, do not make the restaurant futuristic, do not turn the kitchen into a fantasy castle, do not add explosions, do not add unrelated magic effects, do not change the elderly lady chef into a young woman, do not change her glasses or gray-blue hair, do not change the male chef’s mustache or tall chef hat, do not make the male chef win, do not make the lady chef lose control, do not use chaotic unreadable motion, do not overuse camera shake, do not use extreme zooms, do not distort faces, do not deform hands, do not create extra limbs, do not make food look rotten, do not make the dish messy, do not break the final comedic ending.
⬆️ Выше — продолжение промпта для Seedance 2.0.

Да, промпты настолько большие. Но иначе не получится результат на уровне Голливуда.

А чтобы сделать генерацию ещё выгоднее, введите промокод TG15 в личном кабинете. Вы получите +15% к пополнению 🎁
This media is not supported in your browser
VIEW IN TELEGRAM
Поцелуй на 1 МЛН просмотров

Тренд с эфиром спортивных событий не стоит на месте. Вот новое прочтение тренда и промпт под стать:

Style: Hyper-realistic live NBA broadcast footage, authentic televised sports coverage, realistic arena lighting, telephoto broadcast lens compression, shallow depth of field, imperfect live-camera framing, natural crowd reactions, realistic skin textures, subtle compression artifacts, slight interlacing grain, live TV color grading.

Duration: 14 seconds
Aspect Ratio: 16:9
Camera: Fixed live broadcast camera only, no cinematic cuts, no camera movement.

IMPORTANT:
The scene must feel EXACTLY like a real NBA live TV Kiss Cam broadcast.

The woman:
— unbelievably attractive Korean woman in her 20s
— long black hair
— flawless skin
— elegant glamorous makeup
— luxurious aura
— sophisticated red low-cut summer outfit
— subtle jewelry
— natural smile and realistic reactions

The man:
— overweight unattractive man
— casual black jacket
— glasses
— awkward but wholesome energy

IMPORTANT AUDIO RULE:
ABSOLUTELY NO SOUND OR DIALOGUE FROM THE COUPLE.
The couple never speaks.
Only arena crowd cheering and distant stadium ambience.

Broadcast details:
— realistic NBA scoreboard overlay
— sports network watermark
— live broadcast graphics
— crowd in background
— realistic arena LEDs
— slight motion blur
— realistic televised image quality

[00:00-00:03]
Fixed Kiss Cam shot appears on the arena jumbotron. The woman notices the camera and immediately laughs shyly while casually waving her hand “no” toward the screen. The overweight man beside her looks shocked and confused, then smiles nervously. Crowd begins cheering loudly. No talking from either person.

[00:03-00:06]
The crowd cheers louder encouraging them. The woman smiles warmly while glancing at the man. He looks overwhelmed and frozen in disbelief. The broadcast camera slightly struggles to keep framing centered realistically like a real live operator.

[00:06-00:09]
The woman suddenly grabs the man’s cheeks gently and kisses him. The man becomes completely shocked and overjoyed, eyes wide, almost emotionally frozen. Arena crowd erupts cheering and laughing loudly. Realistic shallow depth of field isolates them from crowd.

[00:09-00:12]
The woman pulls away laughing silently and gives a playful thumbs up directly toward the Kiss Cam. The man still looks stunned and unbelievably happy. Crowd continues cheering loudly while broadcast overlays remain visible. End exactly like a real televised NBA Kiss Cam moment.
[00:12-00:14] The man looks in awe and surprise to the camera , the woman smiles with shyness.
Audio:
ONLY loud NBA arena ambience, crowd cheering, distant announcer echo, stadium reactions, sneaker squeaks, arena music muffled in background.
ABSOLUTELY NO dialogue, whispering, laughing audio, or vocal sounds from the couple.

Negative prompts:
No cinematic movie look, no dialogue, no talking, no subtitles, no unrealistic beauty filter, no anime style, no perfect framing, no dramatic camera movement, no unrealistic reactions, no extra people interacting directly with camera.