AI Image Generator

INVARIANTS TO LOCK
- Vertical 9:16 product-announcement Reel for Nano Banana 2.
- Visual language is bold, fashion-editorial, and highly graphic.
- Main recurring character is a young adult man in a red head covering and transparent visor/goggle apparatus, sometimes holding curved yellow bananas near his face.
- Secondary campaign examples include a glamorous woman in superhero-like styling and a pink-suited masked character in a red spider mask working at a pink desk or moving through a city.
- Text overlays drive the story: Google just dropped Nano Banana 2, connected to the internet, web + image search, no references, no uploads, fast campaign visual generation.
- Tone is launch-hype with concrete capability claims.

SHOTLIST
1. [00:00-00:06] Open on the red-capped goggle-wearing character in a studio-like portrait frame while bold white text announces Google just dropped Nano Banana 2.
2. [00:06-00:12] Rapid text beats emphasize that this is the best AI image generator yet, using close portrait crops and minimalist black title cards.
3. [00:12-00:18] Cut to campaign-style examples: a stylish woman in bold fashion-superhero styling pointing at camera, then a pink-suited red-masked figure in a monochrome office setup.
4. [00:18-00:26] Show the pink-suited masked character in multiple scenarios, including desk scenes and dynamic motion, while text explains built-in internet, web, and image search understanding.
5. [00:26-00:37] Finish with side-by-side or sequential campaign visuals that imply the model already knows what objects, people, and products look like, ending on a CTA to comment banana or banana2.

STYLE BIBLE
Visual style: launch trailer for an AI image model, fashion-adjacent campaign visuals, graphic typography over portraiture.
Camera signature: mostly static or lightly animated portrait images, intercut with fast text cards and alternate character shots.
Lighting signature: clean studio light on portraits, candy-color campaign scenes, strong red and pink palette accents.
Grade signature: high saturation, smooth skin, sharp typography, commercial polish.
Speech style: punchy announcement cadence, short lines, product-hype with specific feature claims.

MASTER PROMPT
GLOBAL LOCK: Create a vertical launch-announcement Reel for an AI image model called Nano Banana 2. Build the visual language around bold white text, high-fashion character portraits, and candy-colored campaign scenes. Keep a recurring eccentric studio character with a red cap or hood, transparent visor over the eyes, and bananas held like props near the face. Intercut this with campaign examples of a pink-suited masked figure in spider-like red headgear and a glamorous woman in a comic-book-meets-fashion look. Use the visuals to support the claim that the model can search the web, understand reference-free prompts, and generate full campaign imagery in seconds.

[00:00-00:05] Begin on the visor-wearing red-capped character in a centered portrait frame while text lands word by word: Google just dropped Nano Banana 2.

[00:05-00:10] Use close crops, black title cards, and precise portrait stills to stress that this is the best AI image generator yet. Keep the graphic pacing sharp and premium.

[00:10-00:16] Introduce stylized campaign examples: a confident woman in editorial comic-book styling and a red-masked figure in a pale pink suit working at a matching desk. These shots should feel like instantly generated ad images.

[00:16-00:24] Continue with multiple scenes of the pink-suited masked figure in different setups while text explains built-in web and image search, with no references and no uploads needed.

[00:24-00:31] Show comparative or serial visuals implying the model already knows what shoes, people, and branded objects should look like. Keep the examples punchy and campaign-ready.

[00:31-00:37] End on the strongest studio portrait or fashion visual with a CTA telling viewers to comment banana2 or banana for access.

NEGATIVE PROMPT
Do not drift the signature red/pink visual system, and do not let the campaign examples become generic stock scenes. Avoid muddy typography, weak fashion styling, poor face consistency, or random internet-search metaphors that are not visually tied to premium generated images. Keep the reel feeling like a real launch creative.

SPEECH PACK
[00:00-00:12] Speaker A. Meaning: Google released Nano Banana 2 and it is the best image generator so far. Delivery: emphatic, launch-style.
TAKE_A: “Google just dropped Nano Banana 2, and this is without a doubt the best AI image generator yet.”
TAKE_B: “Nano Banana 2 is here, and the jump in capability is obvious immediately.”
TAKE_C: “This is the kind of update that changes the standard for AI image generation.”

[00:12-00:27] Speaker A. Meaning: it is connected to the internet and can search web and images without references. Delivery: rapid capability breakdown.
TAKE_A: “It is connected to the internet, with web and image search built in, so it already knows what things look like.”
TAKE_B: “No references, no uploads, you type the prompt, it searches, finds the object, and builds the shot.”
TAKE_C: “The big unlock is context: it can understand what you mean without you spoon-feeding it references.”

[00:27-00:37] Speaker A. Meaning: it is fast, campaign-grade, and available in Higgsfield. Delivery: CTA close.
TAKE_A: “This is pro-level quality at flash speed, now live inside Higgsfield, so comment banana2 if you want access.”
TAKE_B: “It is blink-and-it’s-done fast, and if you want access, comment banana or banana2.”
TAKE_C: “Comment banana below and I will send access.”

A vertical talking-head tutorial reel hosted by a young white male creator seated against a solid warm orange studio backdrop. Large kinetic captions introduce a test of multiple AI image and video tools for generating professional-looking avatars. The edit alternates between direct-to-camera explanation, moody retro-tech B-roll of the host at a vintage CRT computer in a dim teal-and-amber room, stylized example portraits arranged in tiled grids, and cinematic concept scenes featuring human characters, analog screens, and fashion-editorial lighting. One standout shot shows a television-headed figure standing beside a woman in a patterned dress, labeled “Midjourney.” Other segments show portrait matrices and tool comparisons, with the overall visual language leaning cinematic, grainy, nostalgic, and premium rather than clean SaaS tutorial aesthetics.

[Subject] A surreal uniformed officer standing in a sunlit institutional hallway, with a dark navy police-style shirt, metallic chest badges, and a calm body posture, while the head transforms into a flowing pale liquid ribbon that stretches sideways and resolves into two connected face forms. [Environment] A blue-toned corridor with tiled lower walls, long ceiling light fixtures, large windows on the right, warm daylight entering from outside, and a metal railing cutting across the foreground, evoking a school, hospital, or civic-building hallway. [Composition/Camera] Vertical mid-body portrait, camera positioned slightly below face level, subject centered but with the liquid head distortion extending dramatically to the right, foreground railing adding depth, corridor perspective lines guiding the eye toward the surreal deformation. [Lighting] Soft daylight mixed with cool interior ambient light, gentle reflections on the floor and badges, even illumination across the uniform, pale luminous highlights on the liquid face ribbon, warm window light balancing the cool hallway palette. [Style/Rendering] Surreal concept illustration with painterly anime-influenced realism, dreamlike institutional atmosphere, clean line structure, soft watercolor-like shading, uncanny visual metaphor, polished poster presentation. [Detail constraints] Preserve the officer uniform as grounded and readable, keep the liquid head transformation smooth and ribbon-like rather than gory, maintain the corridor perspective and windows as contextual anchors, ensure the surreal effect feels uncanny but elegant, and avoid horror clutter or excessive visual noise.

Negative prompt: gore, blood, horror splatter, zombie effect, messy distortion, extra limbs, broken anatomy, low detail hallway, cluttered background, harsh shadows, warped badges, muddy colors, low-resolution painterly blur, duplicated faces without flow connection

Suggested parameters: aspect ratio 2:3, stylize medium, high detail, surreal realism, painterly editorial mood, clean uncanny atmosphere

Delta prompt strategy:
1. If the surreal effect feels weak, increase the pale liquid ribbon stretching the head into two connected face forms.
2. If the image turns horror-heavy, remove gore and keep the transformation smooth, clean, and dreamlike.
3. If the hallway loses clarity, reinforce blue walls, windows, ceiling fixtures, and linear perspective.
4. If the uniform becomes generic, sharpen the police-style shirt, chest badges, and duty-belt details.
5. If the composition feels static, preserve the forward torso while letting the head distortion sweep laterally across the frame.
6. If colors become muddy, separate cool blue interior tones from warm daylight near the windows.
7. If the surreal ribbon lacks elegance, smooth the edges and create graceful flowing curvature between the faces.
8. If the image reads too realistic, add subtle painterly softness while preserving strong structural drawing.
9. If the foreground feels empty, keep the metal railing to anchor depth and realism.
10. If the mood becomes too literal, emphasize uncanny metaphor and poetic visual distortion over narrative explanation.

GLOBAL LOCK: Retro sci-fi action-drama illustrated as painterly cinematic concept art with consistent late-20th-century dystopian thriller energy. Keep the main human lead as a white-presenting adult male in his 30s to early 40s with fair skin, strong jawline, short brown hair styled upward, athletic build, and a stern, protective posture. Keep the teenage boy slim, youthful, and slightly awkward, the red-haired mother tough and alert, the chrome humanoid machine perfectly metallic and expressionless, and the police officer as a surreal shape-shifting impostor whose face splits into a white liquid duplicate floating beside his head. Preserve the American Southwest setting with bars, alleys, concrete flood channels, desert highways, industrial firelight, institutional blue hallways, and motel-town streets. Maintain warm sunlit exterior tones, cool blue interior fluorescents, bright orange fire glow, silver chrome reflections, moderate painted texture, clean action readability, dramatic close-ups, and a mix of 35mm, 50mm, and 85mm cinematic framing. Speech style is sparse and trailer-like, with one or two short lines per beat, tense cadence, dry close-mic sound, and visible lip sync only where characters are front-facing in close-up.

[00:00-00:04] Night battlefield under a moonlit sky, a chrome endoskeleton warrior stands amid explosions and smoke, full-body wide shot with a low camera angle, burning debris behind it, hard orange backlight from fire, cold blue moon fill, drifting smoke and embers, no speech, only apocalyptic tension.

[00:04-00:08] Interior roadside bar or garage with warm practical lights and Pepsi signage, shirtless muscular man faces a woman at close conversational distance, medium close-up with shallow depth of field, tense eye contact, muted amber color grade, slight camera push-in, no speech or a low murmur that feels interrupted.

[00:08-00:12] Surprised close-up of the same male lead, 85mm portrait lens, eyes wide, a white liquid shape creeps into frame from the right edge, skin rendered with smooth painterly highlights, lips barely part as if about to say something, high lip-sync strictness if any whispered word is included.

[00:12-00:16] Wooden doorway confrontation, a heavy bearded man points a shotgun outward from inside a dim rustic room, reverse-shot structure from the visitor’s perspective, warm tungsten light inside, cooler dusk outside, angry expression, fast emotional escalation, cut sharply on the aiming gesture.

[00:16-00:20] Police officer stands behind a metal railing in a blue institutional corridor, daylight windows on the right, his face splits into a white liquid double that floats off to one side, medium shot and then close-up, uncanny identity distortion locked to the character description, dry fluorescent lighting, no comedy, pure body-horror unease.

[00:20-00:24] The police impostor faces a silver chrome humanoid in a workshop or station-like interior, alternating close two-shots and profile shots, the officer studies the machine while the machine remains unreadable, polished reflections on metal surfaces, low conversational tension, spoken line if used should be controlled, cold, and clipped.

[00:24-00:28] Exterior small-town alley with an ATM and pale morning sunlight, a teenage boy stands with another teen beside a red dirt bike, medium-wide framing, concrete walls and utility lines create depth, naturalistic golden daylight, hesitant body language, short casual dialogue possible with loose teenage cadence.

[00:28-00:32] The boy meets the red-haired mother, then they launch on the bike through a yard and into open road, camera alternates between side tracking and rear chase framing, dust and wind motion emphasized, warm sun, hopeful but urgent energy, no visible speech once the ride begins.

[00:32-00:36] Domestic kitchen interior with the red-haired mother alone, plaid sleeveless shirt, checking space around her with alert suspicion, then transition to a long concrete flood channel where a motorcycle races toward camera. Use medium shots indoors, wide vanishing-point exterior frames outside, bright noon light, pacing accelerates.

[00:36-00:40] A chrome humanoid emerges from a wall of fire in a blazing industrial doorway, centered heroic composition, flames licking around perfect metal anatomy, then cut to the main group lined up together like a defensive unit. Keep the contrast high, metal glossy, and fire glow intense, no speech, only mythic escalation.

[00:40-00:44] Return to the blue hallway where the police impostor’s face peels into a vertical white liquid split, then cut to a car interior crossing desert country with the teenage boy in the back seat and the stern protector driving. Tight close-up on the melting face, then side-profile car shots with golden late-afternoon light, minimal dialogue with deliberate pauses.

[00:44-00:48] Reveal a robotic hand behind glass, a Black male observer studies the mechanical fingers, then another man exposes his own cybernetic arm in a bright interior. Macro mechanical details, articulated joints, cables, metal knuckles, cool clinical light, slow deliberate hand motion, no speech or a single stunned reaction word.

[00:48-00:52] Mechanical hand flexes in close-up, then the police figure charges forward in front of fire, furious and no longer convincingly human. Close, aggressive framing, rapid motion, hot industrial glow, smoke, clenched teeth, voice if present should be forceful and urgent with hard consonants.

[00:52-00:56] Leather-jacketed hero loads a shotgun in a workshop, chest-up framing and insert shots of the weapon, then he appears in full figure, bandolier across his body, locked in battle stance. Use crisp action inserts, fiery orange backlight, metallic set dressing, and a hard determined facial expression.

[00:56-00:58] Final desert tag: a black-clad woman in sunglasses holds a rifle near a rugged vehicle and Joshua trees, medium-wide hero shot in harsh dry sunlight, wind moving clothes slightly, no speech, end on a survivalist future-war note.

NEGATIVE PROMPT: low-detail faces, inconsistent identities, duplicated limbs, broken fingers, warped firearms, unreadable props, incorrect police uniform details, cartoon slapstick tone, muddy chrome reflections, flicker between shots, temporal jitter, random text or logos, floating objects without narrative purpose, soft mushy anatomy, deformed motorcycles, broken perspective, accidental modern smartphones, robotic lip movement, off-timing mouth shapes, slurred dialogue, metallic synthetic voice, harsh sibilance, clipped peaks, pumping compression, over-denoised speech, and mismatched room tone between cuts.

SPEECH PACK:
[00:08-00:12] Speaker A, closest audible: "What the hell is that?" Safe paraphrase: "He sees something impossible entering frame." TAKE_A: shocked, breath catches before "hell". TAKE_B: lower, more controlled disbelief. TAKE_C: whispered panic. Lips visible: yes, high sync.
[00:20-00:24] Speaker B, closest audible: "You are not him." Safe paraphrase: "The officer realizes the machine is not human." TAKE_A: flat and clinical. TAKE_B: suspicious and tense. TAKE_C: almost whispered. Lips visible: partial, medium sync.
[00:24-00:28] Speaker C, closest audible: "Come on, let's go." Safe paraphrase: "The teens move toward the bike." TAKE_A: rushed. TAKE_B: nervous. TAKE_C: urgent whisper. Lips visible: partial, medium sync.
[00:32-00:36] Speaker D, closest audible: "Get inside." Safe paraphrase: "A protective instruction before the chase escalates." TAKE_A: firm. TAKE_B: louder warning. TAKE_C: clipped command. Lips visible: low to medium sync.
[00:40-00:44] Speaker A, closest audible: "He's still behind us." Safe paraphrase: "They realize the threat remains active during the drive." TAKE_A: tense low voice. TAKE_B: controlled urgency. TAKE_C: breathy fear. Lips visible: partial, medium sync.
[00:48-00:52] Speaker B, closest audible: "Run!" Safe paraphrase: "Immediate danger forces escape." TAKE_A: shouted. TAKE_B: raw panic. TAKE_C: hoarse command. Lips visible: yes, high sync.

GLOBAL LOCK: vertical 3:4 Adobe Firefly Boards style promo card, static held frame, red brand treatment over a gloomy downtown city block. Main image shows a tall monolithic concrete tower tinted deep Firefly red, torn open by two vertical cracks, with a masked cyberpunk antihero figure emerging from the fissure. Character design: short white hair, white or silver face mask with dark eye slits, dark tactical armor or jacket, menacing upright posture. Preserve Firefly square 'Fi' logo at top left, bold white headline stacked center reading 'From Idea to Branded Mockup' with a red capsule beneath reading 'in minutes', smaller white subhead explaining how AI-first Firefly Boards help visualize concepts without leaving the flow, lower-left hashtags for Adobe Firefly ambassadors and Firefly Boards, and a small swipe cue at lower right. Rainy traffic, buses, taxis, and pedestrians anchor scale at street level.
[00:00-00:11] Hold on the same branded hero frame throughout with only subtle export shimmer. The red building, cracked facade, cyberpunk figure, overcast clouds, and downtown traffic remain static while the large white headline and red capsule emphasize the message that Firefly Boards turns an idea into a branded mockup in minutes.

GLOBAL LOCK: A vertical prompt-demo social video, approximately 15 seconds, with the upper half showing a locked wide cinematic scene and the lower half displaying a long readable prompt block on black labeled “Prompt.” The scene takes place in an open snow field outside a mountain village in Hokkaido during genuine winter. The environment is flat deep snow, bare black birch trees at the field edge, low village rooftops in the background, heavy overcast sky, and a nearly invisible horizon where pale gray sky merges with pale snow. A Japanese master in dark hakama and gi enters from the right, walking slowly through the snow. A white cat wearing a dark gi stands or crouches alone on the left. Their breath clouds merge in the cold air. The two meet near the center, bow very low and very slowly, then rise already in motion and exchange a brief burst of highly committed martial techniques, immediately absorbed and countered, before separating and returning to stillness. The shot should feel like restrained classic cinema rather than comedy. The prompt text below must remain visible and legible throughout. No dialogue; only implied footstep crunch and winter stillness.

[00:00-00:04] Open on the full locked wide shot of the Hokkaido snow field. The white cat in a dark gi is already visible on the left, small against the pale expanse. From the right, the Japanese master in black hakama walks steadily inward, each step compressing the snow. The prompt block remains visible below.

[00:04-00:07] The two figures approach the center and stop at a respectful distance. The master bows deeply toward the cat; the cat lowers in response. Breath and snowfall remain subtle. The visual tone is grave, ritualistic, and sincere rather than humorous.

[00:07-00:11] As they rise from the bow, they immediately enter a short exchange of martial movement. The master pivots and turns with full commitment while the cat holds ground and reacts within the same locked frame. The choreography is minimal, crisp, and surprising precisely because the camera does not move.

[00:11-00:15] The exchange ends as quickly as it began. The master separates and returns toward his side of the field while the cat remains poised on the left. The snow field, birch line, and village roofs reassert the silence. The prompt text still fills the lower half, completing the prompt-to-output teaching format.

NEGATIVE PROMPT: avoid cartoon cat styling, avoid comedy exaggeration, avoid shaky camera, avoid close-up cuts, avoid bright saturated colors, avoid modern props, avoid broken snow physics, avoid martial-arts costumes changing color, avoid extra villagers, avoid anime rendering, and avoid unreadable prompt text.

GLOBAL LOCK: vertical social post layout demonstrating an AI video prompt, top half shows photoreal first-person domestic-catastrophe sequence, bottom half is a persistent black text card with yellow-white prompt copy and a bright yellow CTA reading 'Send this post!'. Top sequence begins as anonymous first-person POV with tanned forearms in bright blue latex gloves pressing a chrome soap dispenser or faucet at a bathroom sink, then escalates into impossible expanding soap foam filling the sink, hallway, staircase, living room, and eventually flooding outside around a suburban house and neighborhood. Camera language is literal and progression-based, moving from POV bathroom realism to wider interior and exterior coverage. Tone is deadpan, absurd, and photoreal.
[00:00-00:03] First-person POV looking down at blue-gloved hands over a white bathroom sink beneath a mirror, pressing the chrome soap dispenser, ordinary warm-lit bathroom realism. Bottom half already displays the black prompt card explaining the scenario.
[00:03-00:05] Dense white foam erupts and swells rapidly above the sink basin, hands spread apart by the pressure, top image remains centered on the bathroom event while prompt text stays locked below.
[00:05-00:08] Cut to hallway and staircase views as the same white foam mass pushes through the house with unnatural volume and even pressure, filling corridors and rooms, furniture disappearing beneath the expansion, bottom prompt card unchanged.
[00:08-00:12] Wider interior and exterior suburban-house shots show foam bursting out windows and doors, piling around the home in thick rounded clusters under overcast daylight, still framed above the persistent prompt text block.
[00:12-00:15] Final aerial or pulled-back neighborhood-scale view reveals the foam event overtaking surrounding streets and lawns, turning the block white while the bottom prompt card and yellow 'Send this post!' CTA remain visible, ending like a shareable AI prompt concept demo.

GLOBAL LOCK: A vertical social video case-study layout, approximately 15 seconds, where the upper half displays a cinematic AI-generated night scene and the lower half permanently displays the generation prompt as readable yellow or off-white text on a black panel labeled “Prompt.” The video content shows a woman in her 40s with Finnish heritage cues, pale eyes, and blonde hair pulled back, wearing a structured dark grey expedition jacket and dark technical trousers. She climbs a rope ladder on the exterior of a glass skyscraper at night, high above a glowing city grid. The mood is calm, determined, and cinematic rather than action-thriller. Lighting is cool blue night city light with warm office windows inside the tower. Camera alternates between closer views of her on the ladder, wider views showing scale on the building facade, and rooftop shots where she waters a tiny plant growing from a crack in the parapet with a metal watering can. The lower prompt block must remain visible and legible throughout, framing the clip as a prompt-to-video demonstration. No dialogue.

[00:00-00:03] Open with a tighter shot of the woman climbing the rope ladder against the reflective glass skyscraper at night. Her dark expedition jacket, focused upward gaze, and rope grip should feel realistic and controlled. The lower third or lower half shows the label “Prompt” and a dense block of prompt text on black.

[00:03-00:06] Cut wider to reveal the scale of the climb. She is small against the tall glass facade, with illuminated office windows behind her and red aircraft lights or distant city lights punctuating the dark skyline. The prompt text panel remains fixed below, functioning like a live case-study caption.

[00:06-00:10] Transition to rooftop arrival. The woman reaches the top edge and moves toward a parapet with city lights stretching behind her. A metal watering can sits nearby. She remains composed, almost ritualistic, as if this impossible rooftop gardening act is normal to her.

[00:10-00:13] Show the woman kneeling or leaning near the parapet as she lifts the watering can and pours water onto a tiny plant growing from a narrow crack in the rooftop edge. The city below glows softly out of focus. The action should feel intimate and quietly poetic after the large-scale climb.

[00:13-00:15] End on the watering action or the rooftop pause, keeping the prompt text still visible below. The final impression should be that of a complete prompt-engineering showcase: one concise narrative arc visualized clearly, with the source prompt presented as part of the content itself.

NEGATIVE PROMPT: avoid action-movie chaos, avoid broken ladder anatomy, avoid unrealistic rooftop physics, avoid extra characters, avoid unreadable prompt text, avoid modern UI overlays beyond the prompt panel, avoid daytime lighting, avoid wrong wardrobe color, avoid flickering plant scale, avoid melted glass reflections, and avoid generic heroic posing.

GLOBAL LOCK: A vertical 9:16 product-demo style social video focused on Adobe Firefly's Generative Expand feature inside a clean motion-design layout. The main image is a retro-inspired female tennis player with light skin and shoulder-length brown hair under a blue visor, wearing a white polo shirt, navy pleated tennis skirt, bright blue socks, and white sneakers. She stands on a green hard court in the middle of a serve motion, one arm raised with a wooden racket and the other arm extended upward toward a tossed ball. The background design uses a dark charcoal frame with bold white headline text reading “Generative Expand” and a smaller italic subtitle explaining that changing aspect ratios should not mean starting over. The key action is not human motion but the image canvas expanding outward to reveal more of the court and complete the composition.

[00:00-00:03] Open on the initial tighter composition of the tennis player centered inside a bordered layout with dark margins around the image. The bold white heading “Generative Expand” sits large in the upper left, while the subtitle sits along the lower portion. The player is frozen mid-serve and the layout feels like a social ad before state.

[00:03-00:05] Animate the image expansion outward so the canvas widens and more of the green tennis court becomes visible around the player. The motion should feel smooth and product-focused, clearly demonstrating that the AI is extending the frame rather than changing the subject.

[00:05-00:07] Finish on the expanded composition where the tennis ball is visible above the player's raised hand and additional court lines and shadow detail complete the scene. Keep the white text overlays legible and the overall result crisp, fast, and workflow-oriented.

NEGATIVE PROMPT: live-action camera footage, multiple tools shown at once, messy UI clutter, different athlete outfit, low-detail court texture, dramatic emotional acting, noisy background, inconsistent tennis pose, random new props, dark moody sports cinematic look.

SHOT PROMPTS: tennis serve image before expansion; Adobe Firefly Generative Expand social demo; widening canvas reveal on hard court; retro tennis fashion AI image; clean production workflow animation.

SPEECH PACK: No spoken dialogue required. The clip should read like a fast visual proof point inside a social ad or carousel video.

Create a vertical cinematic overhead shot of an ornate fantasy carriage traveling slowly through an old graveyard. The carriage and horses should be entirely styled in soft pastel pink, with the carriage looking richly embellished by knitted textures, floral appliques, tassels, and lace-like fabric details. It should feel like a fairy-tale coach reimagined in textile form, with a rounded canopy, decorative trim, and soft sculptural surfaces. Two matching pale pink horses pull the carriage at a slow steady pace along a dirt path between weathered gravestones.

The cemetery should be realistic but subdued: gray and black headstones, scattered leaves, muted brown earth, and a calm overcast atmosphere. Keep the camera elevated in a high-angle tracking view so the carriage remains the hero object while rows of tombstones frame the route. The mood should be surreal and poetic rather than frightening. Emphasize the contrast between the delicate dreamy carriage and the somber graveyard setting.

Use soft naturalistic lighting with restrained color grading so the pink carriage stands out clearly against the earthy background. Motion should remain smooth and slow, with subtle wheel movement and gentle horse steps. Avoid gore, horror creatures, fog-machine cliches, or aggressive haunted-house styling. The result should feel like a visually subversive fantasy tableau: a romantic pastel carriage procession moving through a cemetery in a calm, elegant, slightly uncanny way.

GLOBAL LOCK: a short vertical surreal cemetery tableau showing a wide field of blush-pink ornate gravestones arranged across muted grass and pale stone paths, each memorial marker decorated with bows, pearls, lace, floral appliques, padded satin textures, and ruffled tulle bases; no people, no vehicles, no text; cool overcast outdoor light with a restrained dusty-pink palette against gray-beige earth tones; the scene should feel like gothic bridal installation art, quiet, elegant, and uncanny, with minimal to no motion, vertical 9:16.

[00:00-00:01] Open on a broad elevated view of the cemetery grid, revealing many pink embellished tombstones spread across the lawn with small white flower bundles at the edges.

[00:01-00:02] Hold the composition so the viewer can read the repeating decorative logic across heart-shaped stones, crosses, rounded headstones, and low rectangular forms, all restyled in blush textile ornament.

[00:02-00:03] Preserve the contrast between the soft bridal material language and the solemn burial-ground layout, with pale stepping stones cutting through the frame and overcast light flattening the shadows.

[00:03-00:04] Let the camera drift almost imperceptibly while the repeated pearls, bows, lace trim, and gathered fabric bases remain the visual focus.

[00:04-00:05] End on the same complete cemetery installation view, emphasizing atmosphere, repetition, and the dreamlike couture treatment of every memorial object.

GLOBAL LOCK: Create a short vertical top-down installation-art video showing a perfectly arranged symmetrical field of pale blush, ivory, and soft pink rococo-gothic objects placed on rough gray concrete ground. The composition reads like a tiny dollhouse cemetery, miniature shrine, or pastel memorial installation rather than a standard interior room. At the center sits an ornate heart-backed rococo chair or throne with elaborate trim and scalloped base. Around it are mirrored rows of small sculptural objects resembling miniature tombstones, altars, cherub-like figures, memorial pedestals, carved plaques, and soft baroque decorative blocks. The whole layout must feel curated, ritualistic, collectible, and highly controlled. The camera is a fixed overhead view with almost no movement beyond a subtle stabilization drift or tiny push. Lighting is soft natural or diffuse overcast daylight, preserving surface detail and pale tonal subtlety. No humans, no dialogue, no text, no UI.

[00:00-00:02] Open on the full installation from a direct top-down view. The heart-backed central chair anchors the arrangement, while miniature pale objects form mirrored rows above, beside, and below it. The gray concrete floor should remain visible between objects, adding a stark industrial contrast against the delicate pink-ivory ornamentation. Keep the camera still and the symmetry immediately legible.

[00:02-00:05.04] Continue holding the same top-down composition while allowing only a barely perceptible drift or push that helps the viewer study the central throne, the surrounding tombstone-like forms, the little statuary pieces, and the pastel memorial arrangement as a whole. Preserve the uncanny tension between cute rococo softness and funerary symbolism. End without breaking the symmetry or introducing dramatic movement.

NEGATIVE PROMPT: ordinary living room, standard dollhouse furniture set, bright toy-store shelf display, colorful rainbow palette, black gothic horror scene, candles and flames, dramatic fog, humans entering frame, handheld wobble, side-angle camera, missing central heart throne, missing symmetrical arrangement, missing tombstone-like forms, crowded cluttered composition, text overlays, logos, UI graphics, violent action, dark cemetery night scene, realistic full-scale graveyard, ornate gold palace interior, narrative character performance.

SPEECH PACK:
[00:00-00:05.04]
TAKE_A: [silent] no dialogue, quiet ambient stillness only
TAKE_B: [silent] no spoken words, faint outdoor hush if any
TAKE_C: [silent] static installation presentation, no voice, no lip-sync

GLOBAL LOCK: Vertical 9:16 surreal exterior video showing a hearse transformed into an extravagant pastel-pink lolita object, parked in a quiet cemetery under an overcast sky. Keep the same full-vehicle side view throughout: a white hearse base almost completely covered in layers of blush pink ruffles, lace, bows, rosettes, scalloped fabric drapery, and ornamental frills, turning the vehicle into something between a funeral car, wedding carriage, and gothic dollhouse confection. The setting must remain a real cemetery with gravestones, bare winter trees, muted grass, and gray cold daylight. Preserve the emotional contradiction: death rendered soft, cute, decorative, and disarming. No people, no speech, no text overlays, no dramatic motion; only subtle camera drift and a still observational tone.

[00:00-00:01] Reveal the full pastel hearse parked on a cemetery road, framed from the side. The vehicle is densely wrapped in blush pink lace, bows, gathered fabric, and layered ruffles. Gravestones and bare trees in the background establish a sober real-world setting.

[00:01-00:02] Hold the same composition with a slight camera drift. The decorative surface details become clearer: window drapery, rosettes, scalloped trims, and plush textile volume covering nearly every contour of the hearse. Lighting remains flat and wintry, making the pink embellishment feel even stranger.

[00:02-00:03] Preserve the quiet cemetery atmosphere. The hearse should read as immediately recognizable in function yet visually softened into a lolita fantasy object. Keep the background muted and realistic so the contradiction remains the main event.

[00:03-00:04] Continue the calm exterior study. The lace and ruffle density should feel tactile and excessive, while the hearse silhouette remains legible underneath. No people appear, reinforcing the stillness and unease of the scene.

[00:04-00:05] End on the clearest hero hold of the pink lolita hearse in the graveyard, balancing softness and mortality in one image. Keep the pastel texture rich, the cemetery naturalistic, and the overall mood unsettling through mismatch rather than horror.

Create a vertical close-up video of a tarot card that looks entirely handmade from knitted yarn and soft textile stitching. The card should fill most of the frame and depict a pastel interpretation of The Hanged Man: a small skeleton figure hanging upside down from a horizontal branch, surrounded by a decorative stitched border. Use a gentle handmade aesthetic rather than horror. The entire image should appear woven or embroidered, with visible knit loops, soft thread texture, and slightly fuzzy yarn edges. Keep the palette muted and dreamy, using dusty pink, pale lavender, faded green, and cream tones.

Place the Roman numeral XII above the title area and include the words THE HANGED MAN in a stitched, readable all-caps style along the lower portion of the card. The composition should stay centered and symmetrical like a tarot illustration, with the branch near the top, leafy hanging vines on either side, and the upside-down skeleton occupying the middle. The motion should be minimal: just a slight camera hold or faint parallax that gives life to the card while preserving the feeling of a static textile artwork.

Avoid realistic gore, blood, decay, or any harsh horror cues. The skeleton should feel symbolic and stylized, more like folk art or whimsical fiber illustration than a graphic death image. Use soft, even lighting and a clean neutral background so the tactile knitted-card effect remains the star. The final result should read as a surreal handcrafted tarot card animation with cozy fabric texture and clear visual focus on the Hanged Man motif.

GLOBAL LOCK: Vertical 9:16 surreal exterior video of an ornate cemetery mausoleum transformed into a blush-pink couture architecture object, viewed front-on in a real graveyard under cold overcast daylight. Keep the same chapel-like stone structure throughout, but cover it in layers of pastel pink drapery, bows, gathered fabric, lace, rosettes, scalloped trims, and bridal-lolita ornament so the building feels like a funerary monument wrapped in soft ceremonial fashion. The surrounding cemetery must remain realistic: weathered gravestones, stone pathways, bare winter trees, muted gray-blue sky, and damp subdued atmosphere. Preserve the emotional contradiction of mortality rendered gentle and decorative. No people, no speech, no text overlays, only slight camera drift and quiet stillness.

[00:00-00:01] Reveal the full front view of the pink couture mausoleum in the cemetery. The structure has pointed gothic-ecclesiastical lines, but almost every surface is softened by blush drapery, bows, and lace-like ornament. Gravestones in the foreground anchor the scene in a real burial setting.

[00:01-00:02] Hold the composition with minimal camera drift. The building’s fabric-wrapped surfaces, scalloped decorations, and gathered embellishments become clearer, while the cold gray environment makes the pink treatment feel more uncanny.

[00:02-00:03] Preserve the same solemn cemetery atmosphere. The mausoleum should remain readable as funerary architecture, even as its details drift into wedding-cake, bridal, and lolita visual language. Keep the surrounding stones and trees muted and naturalistic.

[00:03-00:04] Continue the calm study of the structure. The tension should come from mismatch, not action: a monument to death presented as soft ceremonial fantasy architecture. No people appear, reinforcing stillness and emotional distance.

[00:04-00:05] End on the clearest hero hold of the pink mausoleum, balancing graveyard realism with elaborate pastel ornament. Keep the building centered, the cemetery bleak, and the decorative surfaces rich enough to feel tactile and excessive.

GLOBAL LOCK: A vertical editorial video set in a real cemetery, focused on a cluster of ornate pastel-pink funerary sculptures and tombstone-like monuments staged against gray stone graves and dry earth. The objects must feel highly decorative and rococo-inspired, with shell motifs, bows, ribbon carvings, flourishes, and soft candy-colored surface treatment that contrasts with the sober graveyard surroundings. Keep the composition as a frontal-to-slightly elevated view, with muted daylight under overcast or flat natural sky, desaturated gray-brown cemetery tones in the background, and the pink monuments as the dominant visual subject. Preserve a surreal-but-gentle mood, no people, no spoken audio, and include a bold magazine-style text overlay block at the bottom reading like an editorial feature tease.

[00:00–00:01]
Open on a static or barely drifting vertical frame of a cemetery plot where several elaborate pastel-pink tomb monuments sit clustered together. The center structure is the largest, flanked by shell-like and ribboned decorative pieces. The background shows ordinary gray stone headstones, bare ground, and sparse wintery branches, making the candy-pink funerary objects feel uncanny by contrast. A lower-third editorial text overlay is already present near the bottom of the frame.

[00:01–00:02]
The camera holds or drifts forward only minimally, allowing the carved bows, shell ridges, scalloped edges, and ornamental relief work to become easier to read. The cemetery remains plainly real: dirt paths, dark stone, and weathered markers sit behind the soft pink structures. Keep the lighting flat and documentary-natural, not theatrical, so the surrealism comes from the object design rather than dramatic cinematography.

[00:02–00:03]
The frame continues to emphasize the contradiction between softness and mortality. The pink monuments look almost confectionary or dollhouse-like, while the environment remains a practical graveyard. The lower overlay text should stay legible and magazine-like, with a small brand tag and a main line about visual subversion, plus a swipe prompt. No subject movement occurs beyond tiny camera stabilization drift.

[00:03–00:04]
Maintain the same composition and mood, letting the eye move between the central pink memorial, the side shell-shaped pieces, and the gray headstones behind. The color treatment should preserve low-saturation earth tones in the background while keeping the pink objects softly luminous. The contrast must feel conceptual, not cartoonish.

[00:04–00:05]
End on the same eerie but gentle tableau: rococo pastel death imagery placed calmly inside an ordinary cemetery. The text overlay remains in place until the end. No cuts, no character entrance, no weather effects, and no audio-driven events. The clip should read like a short magazine feature intro on visual subversion through stylized symbols of death.

NEGATIVE PROMPT: gothic black cemetery styling, horror monsters, fog machines, dramatic blue moonlight, blood, gore, zombies, people entering frame, moving animals, neon colors, candy shop setting, text glitches, broken lettering, camera shake, warped grave geometry, low-detail carvings, extra monuments appearing, sunny golden hour, lush green grass, interior mausoleum, speech, narration, screaming, subtitles beyond the designed editorial overlay, melted rococo details, cartoon rendering.

SHOT PROMPTS:
Shot 1: vertical cemetery editorial frame with a cluster of ornate pastel-pink rococo tomb monuments against gray graves and dry earth, bold magazine-style lower-third text overlay present.
Shot 2: subtle drift or micro push-in revealing shell motifs, bows, scalloped edges, and decorative reliefs while the realistic cemetery background stays muted.
Shot 3: hold on the contrast between gentle pink memorial sculpture and the ordinary graveyard setting, keeping the bottom editorial text readable.
Shot 4: steady conceptual tableau with desaturated background and softly luminous pink funerary forms, no people and no action.
Shot 5: final still hold on the surreal cemetery composition with the overlay intact.

SPEECH PACK:
Speaker count: 0.
On-camera speech: none.
Transcript: none.
Lip-sync requirement: none.
Audio intention: no dialogue or narration; if any sound exists, keep it to neutral outdoor ambience only.
Delivery takes: not applicable because the reference is speech-free.
Overlay note: preserve a designed editorial lower-third text treatment as part of the visual reference, but do not generate additional captions beyond that intended magazine teaser block.