PICKS

I made this 30-second clip using @higgsfield_ai Created with Seedance 2.5. No reference images. No animators. Just prompts Prompt: [0:00–0:03] VISUAL: PIP, a small scrappy inventor-kid (10, goggles, patchwork jacket, mechanical glove on one hand), weaves through the crowded sky-market carrying a glowing energy core wrapped in cloth, glancing over her shoulder. PIP (muttering to herself): "Just get to the turbine. Don't look back. Don't look—" SFX: bustling market chatter, wind chimes, distant airship engines. [0:03–0:06] VISUAL: A massive armored SKY-GUARDIAN robot (18ft tall, clunky but powerful, glowing red visor) smashes through a market stall behind her, scanning the crowd. GUARDIAN (booming, robotic): "RETURN THE CORE. THIS IS YOUR ONLY WARNING." SFX: stall wood splintering, heavy mechanical footsteps, crowd gasping. [0:06–0:10] VISUAL: Pip sprints onto a rope bridge, glances back — the Guardian rips the bridge's support post clean out of the platform, sending the whole bridge swinging wildly. PIP (yelping, gripping the rope): "Okay THAT one I didn't plan for!" SFX: rope creaking, wood cracking, wind howling. [0:10–0:15] VISUAL: ULTRA-FAST ACTION SEQUENCE. Pip swings across on the collapsing bridge, activates her mechanical glove — it fires a grappling hook onto a passing airship, yanking her clear just as the Guardian's fist smashes the empty bridge to splinters. SFX: grapple-hook fire, metal clank, whoosh of swinging, massive fist impact. [0:15–0:19] VISUAL: Airship pilot (an old grizzled sky-trader, unbothered) glances at Pip now hanging off his ship's railing. PILOT (deadpan, sipping tea): "You're the kid they're after." PIP (still dangling): "Depends who's asking!" SFX: airship engine hum, tea cup clink, Pip's boots scrambling for grip. [0:19–0:23] VISUAL: The Guardian leaps from platform to platform, closing the gap, and grabs the airship's tail fin. HEAVY VFX: the ship lurches violently, energy-balloon ropes snapping, sparks and steam bursting from the engine. SFX: metal groaning, snapping ropes, steam hiss, alarmed pilot shout. [0:23–0:26] VISUAL: Pip climbs to the ship's engine core, yanks a lever, redirecting the storm turbine's built-up energy straight into the Guardian's grip. MASSIVE VFX ENERGY BURST — blue lightning arcs wrap the Guardian, throwing it off the ship in a shower of sparks, tumbling down through the clouds below. PIP (shouting, triumphant): "Static shock — my favorite trick!" SFX: massive electric surge, thunder-crack, Guardian's distant metallic crash fading into clouds. [0:26–0:30] VISUAL: Pip slumps back against the railing, soot-streaked, grinning, hugging the core to her chest as the airship steadies and the sunset glows gold across the sky-city behind her. The old pilot glances over, impressed despite himself. PILOT (small smile): "...Not bad, kid." PIP (catching her breath, grinning): "Told you it depends who's asking." SFX: airship engine settling into calm hum, wind, distant sky-city ambience, warm final music sting.

Made with seedance 2.5 in 1080p Duration: 30s | 16:9 | Ultra-photorealistic live-action medieval war | Single continuous shot | Native Korean audio REFERENCE: Use @img1 only for the woman’s exact face, long black hair, body proportions, and worn silver-steel armor with black base suit, leather belt/bandolier. Keep everything field-worn: scratches, dents, dust, worn leather. 00–05s — SILENCE: Extreme close-up of her face on horseback. Shallow depth of field hides the massive army behind her. Wind moves her hair. Controlled breathing, subtle horse movement, leather and armor creaks. She stares toward the enemy, never at camera. 05–09s — COMMAND: She inhales, then forcefully shouts in Korean: “자아아아아—— 드가자아아아아!!!!” Natural throat/chest engagement and outdoor battlefield reverb. Camera stays tight on her face. 09–20s — CHARGE & REVEAL: Horse reacts first, then pushes forward naturally. Hooves strike dirt, creating dust only on impact. Camera reacts slightly late and physically retreats, gradually revealing her horse, nearby riders, then hundreds of cavalry. No digital zoom or cuts. Horses move independently with realistic weight and timing. 20–30s — MASSIVE CHARGE: Camera continues retreating into a wide battlefield view. Hundreds of cavalry surge forward like a physical wave. Dust builds naturally, banners and armor move with momentum. She remains centered at the front. Enemy line becomes clearer but remains distant. No combat or contact. At 29.5s, just before impact, hard cut to black. CAMERA: Real human-operated movement, subtle handheld imperfection, natural inertia, focus breathing, anamorphic character, physically motivated retreat. No drone, orbit, 360°, slow motion, or digital zoom. LIGHT: Overcast late afternoon, cold diffuse light, muted steel, black fabric, brown leather, dusty earth. No chrome, glow, HDR, or lens flare. AUDIO: Wind and horse breathing → command → individual hoofbeats → hundreds of hooves, armor, leather, banners, dust and ground vibration. No music. STRICT: Same woman, face, hair, armor, horse, battlefield, lighting, weather, Korean dialogue, and central position throughout. NEGATIVE: No face drift, wardrobe/armor changes, extra limbs, duplicate riders, CGI horses, synchronized movement, instant gallop, magic, weapons raised, arrows, explosions, fire, combat, subtitles, logos, watermark, fantasy effects.

I've used 2 image refs as the only reference for this one (both made on MJ), along with this prompt: Hyper-kinetic desert combat spectacle, extreme high-intensity CGI action, thick volumetric cinematic haze and golden-hour dust storms, aggressive anamorphic 35mm lens with heavy motion blur, relentless handheld tracking and whip pans, deep shadows, heat distortion, ultra-realistic particle systems. Use the first provided image as exact reference for the ethereal dark reaper silhouette, flowing tattered robes, skull-scythe and arid desert environment. Use the second multi-view character sheet as exact appearance lock for the muscular tribal warrior (mask, red cape, yellow tattered bone-adorned skirt, dreadlocks, scarred body, barefoot). 0-4s: [Extreme Wide Crash-In] Instant violent collision. Warrior explodes forward through dense dust cloud, red cape whipping. Reaper swings massive scythe in wide horizontal arc. Impact detonates shockwave of sand and embers. No pause. 4-8s: [Dynamic Low Tracking] Rapid exchange. Warrior ducks under scythe, delivers spinning elbow to reaper’s ribs, then upward knee. Reaper phases partially, reforms mid-strike, counters with spectral claw rake across warrior’s chest. Fabric and ash particles explode outward. 8-13s: [Orbiting Medium + Dutch Angle] Acrobatic mid-air clash. Warrior vaults off scythe shaft, red cape fully extended, lands dual-fist hammer blow to reaper’s hood. Reaper retaliates with vertical scythe slam that cracks the ground; warrior rolls, rises into spinning back-kick that shreds more robe into black ribbons. 13-18s: [Handheld Close Combat Frenzy] Brutal ground scramble. Warrior grapples the scythe pole, headbutts the mask, then rips free and unleashes three distinct strikes: upward palm, horizontal elbow, downward axe-kick. Reaper absorbs then unleashes tendril-whip of shadow that lashes warrior’s back, drawing blood and dust. 18-23s: [Slow-Motion Impact Montage] Escalating power. Warrior catches scythe mid-swing, twists, drives knee into reaper’s midsection twice, then spinning heel kick that sends the figure airborne. Reaper crashes down in explosion of ash and sand, robes half-dissolved. Warrior advances without mercy. 23-28s: [Wide to Tight Push-In] Finishing sequence. Warrior seizes the collapsing reaper by the remaining hood, delivers a savage rising uppercut followed by a final two-handed overhead smash. Reaper’s entire form shatters into swirling black ash and ember particles that scatter across the desert floor. Warrior stands over the dissipating cloud, breathing hard. 28-30s: [Hard Cut Reverse Extreme Close-Up] Blurred warrior fills foreground, chest heaving, sweat and dust streaking his mask, mouth open in ragged breath. Sharp deep background: the dark figure slowly reconstitutes from residual ash, scythe reforming in hand, rising and stalking forward with predatory stillness. Photorealistic, ultra-detailed fabric physics and particle simulation, perfect motion blur, volumetric haze at maximum density, coherent anatomy and proportions, stable character consistency, cinematic color grade, movie-level VFX polish.

Game over for Hollywood,because Seedance 2.5 can do all action scenes within low cost budget This scene is made with Seedance 2.5 on polloAI. Prompt: Ultra-photorealistic LIVE ACTION, AAA Hollywood Blockbuster, premium semi-CGI VFX, grounded historical action. ARRI ALEXA 65, IMAX cinematography, Panavision anamorphic. One continuous uninterrupted shot, permanent 40° Dutch Angle. Ground-level IMAX tracking centimeters above fortress streets, Extreme Fast Orbit, aggressive crash zoom, violent whip pan, FPV pursuit, reverse tracking, low-angle pursuit, extreme macro eyes/sword grip/boots/weapon impacts/facial expressions. Hyper slow motion ONLY 0.2–0.35 sec during decisive impacts. Battlefield NEVER empty; dozens of elite Templar warriors remain visible. No static camera, gameplay framing, or music. LIVE ACTION CHARACTER LOCK @Image1 uses weapon from @Image2. A shadow assassin wearing a dark charcoal tattered cloak, deep hood, leather belts, red sash, dark combat boots, carrying a long curved sword. His face is partially shadowed by the hood, but his calm, focused eyes are clearly visible. Long flowing dark cloak with realistic fabric folds. Preserve Image 1 exactly: face, hairstyle, body proportions, skin, clothing, cloak, identity, natural eye color. COMBAT STYLE Extremely refined and aggressive high-speed sword style fused with Taijutsu. Sword technique combines precise blade control, rapid parries, deflections, and devastating counter-strikes. Exceptional footwork, hip/torso rotation, shoulder evasion, balance, kicks, rapid pivots, low sweeps, spinning heel/side/roundhouse kicks, jumping knees, foot traps and reverse-grip transitions. No wasted movement; every defense immediately becomes offense. BLINK SPEED Image 1 has overwhelming superhuman burst acceleration. At maximum speed, almost impossible for camera/human eye to track. Each burst leaves short black translucent afterimage, compressed air, dust displacement, atmospheric distortion. Can cross several meters in fractions of a second and reappear behind/beside/above/between opponents through visible physical movement. Camera frequently overshoots, then violently whip-pans to reacquire. LOCATION Massive original Alhambra fortress complex, 15th century Granada, Spain. Stone fortress walls, tiled rooftops, narrow stone alleys, wooden balconies, courtyards with fountains, towers, warm Mediterranean sunlight casting long golden shadows. No recognizable copyrighted symbols or exact landmark replicas. Main battle tears through the broad central avenue and rooftop levels. ENEMY WARRIORS Dozens of elite Templar knights in grounded 15th-century armor: chainmail, surcoats, helmets, cloaks, and swords. Coordinated acrobatics, rooftop pursuit, rapid melee and tactical formations. No recognizable copyrighted uniforms or clan symbols. 00:00–00:03 Ground-level IMAX tracking races through the crowded central avenue as dozens of Templars descend from rooftops, balconies and elevated walkways. One attacks from the side; Image 1 tilts backward, blade missing face by centimeters. Extreme macro calm eyes. Image 1 rotates sword into a one-handed parry — CRACK! Steel clashes. Image 1 redirects attacker downward, steps onto enemy shoulder and launches upward. Mid-air rotation → violent side kick sends another Templar through a wooden market stall. Before landing — BOOM! burst acceleration, black afterimage, violent whip pan. 00:03–00:06 Camera finds Image 1 already sprinting across a sloped tiled rooftop while multiple Templars pursue across neighboring roofs. Low-angle tracking from street below. Throwing blade crosses lens; Image 1 ducks without slowing. Another warrior leaps between rooftops. Image 1 plants one foot on a chimney structure, runs two steps sideways along wall, launches backward over attacker → mid-air sword deflection → immediate spinning heel kick sends Templar through roof tiles. Image 1 lands on roof edge and instantly bursts treme burst

Not from any Hollywood movie but AI Made using Seedance 2.5 Prompt : Young Caucasian woman, chin-length auburn bob haircut, green eyes, wearing a light green short top and white denim shorts, sitting pensively on a sofa inside a large bright suburban Parisian house, summer daytime, sunlight through the windows, young adults dancing and laughing around her holding drinks, loud music implied by energetic crowd movement. She rises from the sofa empty-handed, walks toward a kitchen table lined with alcohol bottles and glasses. Handheld camera follows close behind her shoulder, slightly shaky. She pours herself a glass of champagne, sips it. Camera holds behind her, shaky, close distance. She weaves through the dancing, laughing crowd, some standing, some in groups laughing on sofas, toward a large glass bay window. Camera continues following close behind her, no cut. She steps outside into the garden, a big pool comes into view with young adults swimming and laughing inside it. Camera smoothly rotates from following behind her to facing her. She looks toward the pool with amusement, smiles, sets her champagne glass down on a small table beside a lounge chair. She removes her jersey, places it on the back of the lounge chair, then removes her denim shorts, places them on the lounge chair, revealing a pink swimsuit underneath. Camera holds a medium shot facing her, slight handheld sway. She walks toward the pool, camera repositions behind her as she walks, then she breaks into a run. Handheld camera follows close, shaky, matching her pace. She dives into the pool shouting "take care", landing in the water among other laughing young adults. Camera dives in with her, briefly submerged underwater alongside the swimmers. She resurfaces laughing, joins the laughing adults nearby. Camera is up in the air, revealing a wide aerial view of the house, garden Bright natural summer daylight, warm sunlit color palette, candid documentary energy, slight handheld lens wobble throughout. ..

FEMALE SPY vs. ASSASSIN — 30-SECOND CINEMATIC FIGHT FORMAT: 16:9 widescreen, ultra-realistic cinematic action, premium spy-thriller aesthetic, fast-paced editing, dramatic lighting, dynamic camera movement, realistic choreography. MAIN CHARACTER: A confident female spy in her late 20s, athletic build, sharp features, sleek black tactical suit, black boots, hair tied in a tight ponytail. She is calm, highly skilled, and intimidating. Maintain the exact facial features, eye color, skin tone, hair color, and overall identity from the provided character reference throughout every shot. 0–4 SEC — HOTEL HALLWAY The female spy walks alone through a luxurious dimly lit hotel hallway. She suddenly stops. An assassin appears behind her. She turns instantly as he throws the first punch. 4–9 SEC — HAND-TO-HAND COMBAT Rapid close-quarters combat. She blocks a punch, delivers a powerful elbow strike, ducks under another attack, spins and kicks him into the wall. Fast cuts, whip pans, impact close-ups. 9–14 SEC — GUN DISARM The assassin pulls a handgun. She grabs his wrist before he can fire, twists his arm, strips the gun away, and kicks him backward. The gun slides across the polished floor. 14–19 SEC — KNIFE ATTACK He suddenly pulls a knife and lunges. She narrowly dodges the blade, grabs his arm, redirects the attack, and sends him crashing through a glass table. Glass explodes in slow motion. 19–25 SEC — ELEVATOR FIGHT The spy rushes into an elevator. The assassin follows. The doors close as they continue fighting inside the tight space. She uses the elevator walls for leverage, flips him down, and pins him against the floor. 25–30 SEC — FINAL MOMENT The elevator doors open. She calmly steps out and straightens her jacket. Behind her, the defeated assassin remains on the floor. The doors slowly close. FINAL CLOSE-UP: She looks directly toward the camera with a cold, confident expression. CUT TO BLACK.

He brought a blade… she didn't need hers. Seedance 2.5 in @TopviewAIhq Fight scene prompt inspired by @sophiaparkerr_ [Generation goal] Live-action spy action, one continuous take. Rogue Viper is ambushed in a hotel corridor by a masked ninja, disarms his knife bare-handed, and kicks him off a balcony over a city at dusk. Exactly two people throughout. [Reference roles] Image 1 defines Rogue Viper only: face, amber eyes, long loose black hair, black leather suit, gold cobra buckle, over-knee boots, sheathed katana across her back. All three panels in image 1 are one woman. Her face is visible in every shot and matches the close-up panel. Do not copy the black oval covering the face in the full-body panels, the gray studio backdrop, the panel layout, or the inset buckle image. Image 2 defines the corridor set only: dark green walls, mahogany doors and paneling, blue-and-gold patterned runner carpet, crystal chandeliers, shaded table lamps, warm low light. The katana stays sheathed on her back the whole clip. She never draws it. One sword exists. The ninja has no reference: lean man, matte black garb, wrapped forearms, red lacquered demon mask with gold fangs and small horns. His face is never seen. [Stage 1 - approx 0-6s] Viper walks away from camera down the corridor. She stops. The ninja steps from a doorway behind her, frame right. She turns as his punch arrives and blocks it with her forearm. Camera dollies in behind her at eye level. [Stage 2 - approx 6-13s] Close-quarters exchange mid-corridor. She slips a punch, drives an elbow into the mask, ducks a hook, shoulders him into the paneling. A table lamp topples. He stays on his feet. Camera whip pans with the strikes. [Stage 3 - approx 13-20s] He draws a straight black knife and lunges at her chest. She turns her hips out of the line, traps his wrist in both hands, twists it outward, and the knife drops point-first into the carpet. She stands over it empty-handed. Low medium two-shot. [Stage 4 - approx 20-26s] She drives him backward through tall open glass doors onto a stone balcony. Sky opens behind him, dense city skyline in orange and deep blue dusk. She plants and kicks him in the chest. He goes over the balustrade and out of frame. Camera tracks with them, then stops at the rail. [Stage 5 - approx 26-30s] She holds at the rail one beat, breathing, wind moving her hair. She turns her head to camera and the corner of her mouth lifts into a small smirk. Hold. Cut to black. Restrained, no laugh, no wink, no dialogue. (low bass note swells and cuts) [Maintain consistency] Two people only. Viper's face, hair, suit, and buckle never change. The katana stays sheathed through every strike. The mask stays on the entire clip. One knife, and it stays in the carpet after Stage 3. The fight travels one direction, away from camera, and never reverses. Corridor light warm and low, balcony light cool dusk. No dialogue. No on-screen text, no logos.

MiniMax H3 Fashion Lookbook | AI Outfit Change + Editorial MV Prompt👇 Create a 15-second 16:9 fashion lookbook MV combining rapid outfit changes, magazine editorial graphics, and beat-synced transitions. The video is not narrative-driven. Core structure: Locked camera → Typography reveal → Close-up → Rapid Look switches → Hero Ending Keep the same female character centered in frame throughout the video. Preserve facial identity, body proportions, skin tone, and overall appearance across every outfit. 0.0–1.5s Pure white studio, girl standing naturally in the center. Camera locked. Huge black typography “ONE HIT” rapidly expands behind her. Magazine elements appear on the beat: barcode, scan lines, crosshair, numbers, thin borders, grid graphics. The clean studio instantly becomes a bold fashion magazine poster. 1.5–3.0s Fast camera push-in to a close-up. The girl gently turns toward camera, hair moving naturally, direct eye contact, calm and confident expression. Black-and-white editorial background with subtle film grain. 3.0–4.5s Look 01. Instant outfit change on the beat. Character stays centered and makes a natural fashion pose. Background switches to a completely new magazine layout with oversized headline, barcode and numbering. 4.5–6.0s Look 02. New outfit. Medium shot. Girl takes one confident step forward. Background typography rapidly changes. Black and white remain dominant with stronger red accents. 6.0–7.5s Look 03. New outfit. Girl makes a gentle half-turn, hair swinging naturally. Background changes to red-and-white editorial graphics with huge English typography, numbers, coordinates, scan lines and borders. 7.5–9.0s Look 04. Sporty fashion outfit. Girl leans slightly forward with a relaxed confident smile. Subtle low-angle camera. Large red magazine typography slides into frame on the beat. 9.0–10.5s Look 05. New outfit. Girl moves into a stylish crouching pose. Quick camera push-in. Background layout completely rearranges, creating a new visual impact. 10.5–12.0s Look 06. Girl stands again. Camera pulls back to medium shot. Bold yellow fashion uniform with long blue hair styling. Background returns to black-and-white editorial design with large headline and barcode elements. 12.0–13.5s Look 07. Sharp tailored suit. Girl walks confidently toward camera. Background switches to red-and-black magazine design. New headline, number codes, scan lines and graphic grids appear rapidly. Pacing becomes faster. 13.5–15.0s Final Hero Shot. Girl wears a minimal black fashion look. Arms crossed, standing confidently in the center. Camera settles into a clean medium shot. Background gradually simplifies, leaving only a few premium magazine graphics. Final logo or visual mark appears naturally. Freeze on the final beat. Overall Style Premium Korean fashion lookbook, editorial fashion photography, Y2K magazine design, high-end commercial photography. Black and white base with red accents. Oversized English typography, barcode, scan lines, crosshair, coordinates, numbering, grids, thin borders, subtle film grain. Switch to a new outfit or visual layout every 1–1.5 seconds, synchronized closely with the music beat. Keep the camera relatively controlled. Use quick push-ins, subtle pull-backs, pose changes, graphic transitions, and layout swaps. Character must always remain near the center of frame. Maintain realistic hair movement, fabric physics, natural poses, clean studio lighting, and high-contrast premium photography. Avoid No face drift, character replacement, body morphing, extra limbs, distorted hands, unstable proportions, messy typography, chaotic camera movement, random backgrounds, cartoon style, low-quality CGI, watermark, or unwanted subtitles. Final result should feel like a high-end Korean fashion campaign mixed with a Y2K magazine cover and beat-synced music video. Create with Ima Studio 👉

Seednace 2.5 Prompt 15-sec Seedance prompt — 5 separate shots Style: Authentic early-2000s MiniDV home-video footage. Young Korean woman, 24, long dark hair, minimal makeup, oversized cozy sweatshirt and loose pajama pants, wrapped in a soft blanket. Quiet apartment balcony at night, warm indoor light spilling through the open door, cool blue city light outside. 4:3 MiniDV, handheld imperfections, low-light noise, tape grain/compression, slight autofocus hunting. Natural candid behavior, no cinematic polish. SHOT 1 — 0:00–0:03 Interior wide shot facing the balcony door. She walks toward the open door wrapped tightly in a blanket, sleepy and curious. She slides the door open a little further and steps outside. SHOT 2 — 0:03–0:06 Cut to a balcony-side angle. She takes a few small steps outside, holding the blanket around her shoulders. She looks up at the quiet night sky and takes a deep breath. SHOT 3 — 0:06–0:09 A sudden cold gust blows across the balcony. Her hair moves strongly in the wind and she immediately scrunches her shoulders, pulls the blanket tightly around herself, and makes a funny surprised expression. SHOT 4 — 0:09–0:12 She quickly retreats back through the balcony door, almost stumbling from how fast she moves. She closes the door halfway, still wrapped tightly in the blanket, and looks back outside. SHOT 5 — 0:12–0:15 She slowly peeks her head back around the door and looks toward the camera. She realizes how ridiculous her dramatic retreat was and starts quietly laughing. She shakes her head, smiles, and disappears back inside. End: Balcony remains visible for a brief moment as the door closes. Authentic low-light MiniDV texture, natural handheld movement, subtle exposure adjustment when moving between the warm room and dark balcony. No slow motion, no exaggerated acting.

非クリエイターこそ、Claude Codeに動画生成のプロンプトを書いてもらいましょう😃 げんざい最強の動画生成AIといっても過言ではない「Seedance 2.5」。 長尺の動画を作れる一方で、プロンプトも長くなりがちです。 しかも、Seedance 2.0とはプロンプトの書き方が異なる部分もあるので、自分で毎回最適なプロンプトを作るのは結構大変。 テンプレートを用意してコピペする方法もありますが、それでも毎回調整するのは面倒です。 そこで便利なのが、Claude Codeの「Skill」。 まず、以下のSeedance 2.5のプロンプトガイドをClaude Codeの入力欄に貼ります。 ・https://t.co/BPHSzqNopm ・https://t.co/aUJgXKFRip ・https://t.co/u79zESgiTk いずれもDreaminaやByteDanceが公開している資料なので、Seedance 2.5のプロンプト作成に使う情報としてかなり有用です。 そしてClaude Codeに、 「これらのURLの内容を読み込み、Skillにして。Seedance 2.5のプロンプト生成を依頼する時は、毎回このSkillを使って」 と指示。 以後は動画の内容を伝えるだけで、ガイドに沿ったSeedance 2.5用のプロンプトを自動生成してもらえるようになります。 いまは動画クリエイターだけでなく、 ・企業の広告宣伝担当者 ・企業のPR・広報担当者 ・広告代理店 などでも、AIで動画を作る人が増えています。 すべてを自分の頭と手で担当したら、負担は増える一方。 AIに任せられる部分は、Claude Codeのようなツールにどんどん任せましょう❗️ ※Codexでも同じようなことは可能です そして、人間にしかできない仕事に集中する。 これからのAI活用では、こういう考え方がますます重要になりそうです😃 添付の動画はClaude Codeにプロンプトを書いてもらい、t2vで生成したものです。 Claude Codeに依頼した時のプロンプトはこちら↓ ※Seedance 2.5のプロンプトはリプ欄で紹介 ------------------------------- Seedance 2.5で、15秒の夏らしい清涼飲料水のCMを作りたい。 清涼飲料水の名称は「AOI WATER（アオイ・ウォーター）」 アスペクト比は16:9。 CMに登場するのは架空の日本人若手女優。 さわやかで好感度抜群な若手女優、という設定。 飲めば夏に負けない、フレッシュな美味しさを表現できる、見た人が元気になるような動きのあるCMのプロンプトを書いて。

Higgsfield just dropped the exact prompts behind their best Originals shots👀 Seedance 2.0 on Higgsfield AI Full open-sourced prompts & assets below: (Image1) is the performer — preserve her exact identity: cornrow braids, septum ring, statement earrings, sculptural white designer top, dark indigo denim. (Audio1) is the finished master track — the only audio; no invented music, no new vocals. SHE RAPS THE VOCAL ON CAMERA — PRECISE LIPSYNC IS THE TOP PRIORITY. Mouth articulates every syllable of (Audio1) exactly on time; face visible and sharp through every vocal line, no cutaways mid-word. LIPSYNC MAP: 0.0–0.5 instrumental. 0.5–2.8 "I'm standing on the edge / Say it with your chest / Or keep it on the deck" + "Hey!" 3.2–6.7 "I walk in, whole room gets tense / I don't need luck, I'm the consequence / If you really want to test my intent / Come correct, come correct or get bent" + "Woo!" 7.5–13.7 same hook verbatim second time, escalated. 14.5–15.0 instrumental hold. Music-video route: performance with mirrored echoes. Director thesis: white infinity studio — she raps at the lens while five dancers in white repeat her last pose one beat late, a human delay effect behind her voice. Visual world: white cyclorama infinity studio, seamless floor and walls, one hard fashion key light with clean shadows, subtle floor reflections. 5 female background dancers in all-white utilitarian streetwear, hair slicked, deliberately similar but never identical to her; her indigo denim makes her instantly readable. Palette: white on white, skin tones, indigo. No neon, no particles. Shot flow: 0–2.8s symmetrical wide-to-medium push-in: she raps the opening lines at the apex of a tight wedge, the five echoes frozen in her exact stance behind her. 2.8–3.2s on "Hey!" all five snap chins up in unison. 3.2–6.7s medium: she raps while throwing an angular vogue accent at the end of each line, and the echoes replay that exact accent one beat later, rippling backward through the wedge; camera slowly orbits 45 degrees keeping her mouth front and center; finger to lens on "come correct". 7.5–10s cut on the kick to a chest-up close frame: second hook with doubled intensity, the echoes now a soft-focus rhythmic blur behind her articulation. 10–13.7s the echoes carousel slowly around her while she stands still at center rapping the final lines, camera counter-rotating, her face never leaving focus. 13.7–15s on "Woo!" she freezes arm high; the five echoes freeze in five different mid-move poses around her — she is the only resolved image; micro push-in, hold. Loopable. Performance rules: dominant, stoic, immaculate diction; echoes expressionless and precise, never mouth the words — only she raps. Continuity: same six women, same wardrobe, same white studio. Audio intent: (Audio1) only, her mouth locked to it; faint studio room tone. Quality bar: expensive fashion-campaign rap video, no AI gloss, no glow.

I tested This insane cinematic prompt… on Seedance 2.0 and this feels like a Hollywood trailer 🎬📷 A giant snake attacks New York City… crushing cars, climbing skyscrapers… and wrapping around the tallest tower. This one is pure chaos. The Prompt I used: 16:9 aspect ratio, 15 seconds total duration, 6-shot cinematic sequence, Hollywood blockbuster action trailer style, photorealistic 8K, ultra-detailed textures, dramatic volumetric lighting with sun flares and god rays, realistic physics and motion. Shot 1 (0-3s): Chaotic street-level view in downtown Manhattan New York City on a busy intersection with yellow taxis, black SUVs, panicked pedestrians running and screaming. A massive 200-foot-long prehistoric Titanoboa-style giant snake with dark gray-black scaly skin, glowing orange eyes, and enormous head suddenly bursts out of thick dust clouds, smashing through cars and buildings, debris flying everywhere, high-speed dynamic camera. Shot 2 (3-5s): Low-angle tracking close-up as the giant snake's massive head smashes into a modern glass skyscraper, glass shattering, sparks and fire exploding, orange embers raining down, motion blur on debris, intense destruction. Shot 3 (5-7s): Extreme low-angle worm's-eye view climbing up the snake's huge scaly body wrapped around the skyscraper, sunlight glinting off wet scales with strong lens flares and blue horizontal flares, building reflections visible. Shot 4 (7-9s): Epic wide aerial shot — the enormous snake coils tightly around the top of One World Trade Center (Freedom Tower) in a bright sunny New York City skyline with river and other skyscrapers visible, black smoke rising in the distance, multiple military helicopters flying around it. Shot 5 (9-12s): Dramatic close-up on the snake's menacing head and upper body wrapped around the tower spire, looking down at the chaotic city below, sharp fangs visible, intense orange eyes staring forward. Shot 6 (12-15s): Final epic wide cinematic shot massive fiery explosion erupts at the top of the tower where the snake is coiled, huge orange fireball and thick black smoke billowing outward, snake head roaring, helicopters scattering, intense sun flare and god rays, high-stakes disaster movie climax.

Seedance 2.5 prompt template for a beer commercial below 🍺 👇 High-energy American premium beer commercial in the kinetic visual language of top Spanish beer ads (Águila-style dynamism): extremely fast rhythmic cutting, aggressive camera movement, constant forward energy, product always the absolute hero. Shot on anamorphic cinema lenses with rich contrast, warm golden highlights, deep blacks, subtle film grain, perfect motion blur, and hyper-detailed product beauty. Every frame feels alive, expensive, and addictive. Brand: TALON Logo: bold condensed American industrial sans-serif “TALON” in pure gold, with a sharp geometric falcon talon gripping a single hop cone integrated into the letter “A”. Bottle: tall elegant amber-glass longneck, matte black body label with the gold TALON logo centered, thin gold foil neck collar, embossed talon mark on the shoulder, heavy cold condensation beads, liquid a luminous deep-golden amber with dense creamy white head when poured. 0-2s: [Extreme close-up, rapid whip-pan] Ice explodes as a hand yanks a freezing TALON bottle from a metal cooler. Condensation flies. Instant cut. 2-4s: [Dynamic tracking, low angle] Bottle spins in slow-motion mid-air against a blazing desert sunset highway, gold logo catching the last light. Hard cut. 4-6s: [Handheld kinetic] Young crew in a vintage convertible flying down an empty American desert road at golden hour, wind ripping through hair, laughing. Driver raises a cold TALON without looking away from the road. Cut on the laugh. 6-8s: [Macro beauty, 120fps] Perfect pour: thick golden liquid cascades into a chilled glass, dense foam rising in ultra-slow motion, tiny bubbles racing upward, condensation racing down the glass. Sound of liquid is almost tactile. 8-10s: [Fast montage, 4 cuts in 2 seconds] Rooftop at dusk – city lights ignite. Friends clink bottles hard. Foam overflows. A girl throws her head back laughing. A guy points at the sky. Every cut lands on the bottle or the foam. 10-13s: [Steadicam push through crowd] Night football tailgate, American flags, smoke from grills, pure energy. Someone cracks a TALON, the distinctive sound cuts through the noise. Camera finds the bottle first, then the smiling faces around it. 13-16s: [Ultra-close product sequence] Three rapid macro shots: – condensation drop sliding down the gold logo – thick white head collapsing slowly over the rim – bottle rotating under hard practical light, label flashing gold Hard rhythmic cuts timed like drum hits. 16-19s: [Aerial + ground hybrid] Drone diving over a nighttime desert campfire circle. Friends sit around the flames, bottles in hand. Camera lands on one bottle standing upright in the sand, glowing from the firelight, logo sharp. Instant cut to a hand grabbing it. 19-22s: [Aggressive handheld, pure chaos energy] Bar bathroom mirror selfie energy turns into a full dance-floor explosion. Bottles raised high above the crowd. Liquid flying. Faces pure joy. Camera never stops moving, always finding the TALON label in the chaos. 22-25s: [Slow-motion climax beauty] Single perfect bottle of TALON held against a black background, spinning slowly. Gold logo catches every light. Then the same bottle is slammed down onto a wooden table in real time – foam erupts. Freeze on the eruption for half a second. 25-28s: [Final product hero] Extreme close-up of the open bottle and a full glass side-by-side, both sweating, foam perfect, gold logo glowing. Soft anamorphic flare. The word “TALON” appears in the same gold type over the image. 28-30s: [Hard cut to pure black] Gold TALON logo + talon icon slam on screen with the line: “TALON. Grip the night.” Hold for one beat of pure silence, then cut to black. Cinematic anamorphic look, rich golden color grade, deep contrast, aggressive rhythmic editing, constant kinetic camera energy, hyper-detailed product macro photography, perfect condensation and liquid physics, zero dead frames, pure American freedom energy filtered through European beer-commercial precision, no dialogue, pure visual and emotional momentum. Place your product as reference, tweak the prompt a bit and you are good to go! Hope you find it useful 😎

“I know nothing about her… wild, calm, unbearably alive as if time was stopped for her alone.” This one was planned as a more poetic clothing ad. But now it feels like it could be the beginning of a completely different story💕 Made with Seedance 2.0 on @higgsfield Prompt

Seedance 2.5 Forged in Fire text to video prompt is below 👇 Reality TV bladesmith competition testing segment, multi-camera documentary production, bright industrial workshop lighting, sharp HD broadcast look, practical forge practicals, natural handheld shake on action cams, clean documentary texture, no cinematic film look. 0-4s: [Medium Two-Shot] Martial arts master of clear Asian ancestry in black tactical shirt stands facing the young female bladesmith in her early 20s (work apron, focused). He holds her forged machete, turns the blade once under the lights. Looks her in the eye: “Are you ready?” She answers firmly “Yes.” Quick cut to her determined face, then to three classic ceramic garden gnomes lined up on sturdy stands behind them. 4-7s: [Low-Angle Tracking + Rapid Cuts] Master steps in, raises the machete. Hard cut to side angle as he delivers a powerful overhead chop into the first gnome. Impact explodes in slow motion: ceramic shatters into sharp fragments, paint chips and dust fly outward. Time locks at peak penetration. Camera orbits the frozen blade buried in the broken gnome for one second. Quick insert ECU of the edge biting through the ceramic body. 7-11s: [Dynamic Side Tracking] Action resumes. Master yanks the machete free, immediately follows with a heavy diagonal slash across the second gnome. Slow-motion again: the figure tears apart sideways, large colored pieces and shards scatter. Freeze at maximum destruction. Two rapid cuts: one over-shoulder from behind the master, one low angle under the flying fragments. Orbit holds for half a second on the locked impact. 11-15s: [Handheld Close Action] Third strike — short, vicious downward chop into the third gnome. Slow-mo impact, the ceramic erupts into dozens of pieces, hat and beard flying. Full freeze. Camera pushes in and circles the frozen chaos for one beat. Quick insert of the master’s focused face mid-swing, then back to the locked blade. 15-19s: [Series of Fast Cuts] Time resumes. Master pulls the machete out, turns it under the light. Tight ECU tracks along the edge: a few small nicks and rolled spots visible. Cut to his fingers testing the edge. Cut to the ruined garden gnomes still shedding fragments. Cut to the young woman’s tense expression as she watches. 19-24s: [Medium Close-Up] Master lowers the blade, looks straight at her. Calm, professional delivery: “There are a couple of nicks on the edge… but your weapon, ma’am… it kills.” Hard cut to her face — a small, controlled smile of pride breaks through. 24-28s: [Final ECU Montage] Extreme close-up of the nicked but still lethal edge catching the light. Cut to the shattered remains of the garden gnomes. Final frame holds on the young bladesmith’s quiet satisfaction as the master nods once. Realistic TV broadcast quality, detailed ceramic shatter physics, sharp edge detail, natural motion blur only on moving elements, stable characters, coherent impact timing, no artifacts, high rewatch value. I promise I didn't use any image reference for this one, use it wisely 🫡

Seedance 2.5 on @TapNow_AI TapNow's upscaling is really solid ======================== Prompt: Use @ Emma Asagiri from the uploaded character reference sheet as the exact character reference. Preserve her facial identity, eye color, skin tone, long straight jet-black hair with wispy blunt bangs, body proportions, makeup, and outfit consistently throughout every shot. She wears her sailor-style uniform from the reference sheet: a white puff-sleeve sailor top with a navy collar and white striped trim, a navy neck bow, a navy pleated mini skirt, white knee-high socks, black loafers, a black leather choker, and a white bunny-ear headband. Maintain the same recognizable performer in every scene. Create a 15-second ultra-realistic Y2K rap/pop music video inspired by early-2000s music television aesthetics, filmed with an exaggerated ultra-wide fisheye lens. The visuals should feel energetic, glossy, slightly surreal, and heavily stylized, with dramatic lens distortion, dynamic camera movement, and fast beat-synchronized editing. — Light-Up Reveal: Begin in complete darkness inside a long futuristic black corridor. The rectangular wall lights suddenly switch on one after another, creating a rapid chase effect toward the performer. She stands completely still in silhouette with her arms spread. On the final light hit, the overhead lights illuminate her face as she immediately begins lip-syncing directly toward the camera. — Fisheye Close-Up: Hard cut to an extreme fisheye close-up of her face. She leans aggressively toward the lens, performs confident hand gestures, points directly at the camera, and briefly pushes her palm toward the lens, making her hand appear dramatically oversized from the fisheye distortion. Chrome Room: Cut to a glossy futuristic white-and-chrome room. She performs a playful, confident dance with shoulder movements, small footwork, and sharp hand gestures. The camera smoothly slides sideways at knee height while chrome surfaces create strong reflections and parallax. 6–8s — Graffiti/Dollar Set: Hard cut to an abstract orange-and-lime studio filled with oversized glossy black geometric shapes featuring generic dollar symbols. She performs in profile while lip-syncing, then suddenly turns her head toward the camera on the beat. Her long black hair, navy neck bow, and pleated skirt move naturally with the turn. 8–10s — Dancer Performance: Reveal two anonymous female backup dancers wearing glossy black full-coverage futuristic outfits and dark visor sunglasses. They flank the performer and perform synchronized Y2K choreography while she dominates the center, pointing toward the lens and bouncing confidently to the rhythm. 10–12s — Beam Dome: Cut to a dramatic low-angle shot. She stands on a small pedestal inside a dark futuristic dome as powerful white light beams switch on around her one at a time. She raises both arms and continues performing while the lights create intense silhouettes and lens bloom. — Chrome Macro: Extreme tilted fisheye close-up of half her face against a quilted metallic-silver wall. She gives a playful smirk and delivers a

This time I didn't tag each image, just asked it by text to use them as refs: epic backyard war. Hyper-dynamic cinematic action, aggressive handheld + sweeping crane moves, extreme speed ramps, freeze-frames on impact, dramatic golden-hour backlight, dust and debris in slow-mo, intense heroic angles, pure energy on a suburban scale. Use the provided character turnaround sheets for both boys, the handmade wooden slingshot prop, and the ramshackle treehouse environment plate as strict visual references. Lock exact faces, hair, freckles, clothing, capes, dirt, and prop details. No redesigns. 0-3s: Tense standoff. Rapid intercut extreme close-ups: freckled 9yo boy (messy sandy hair, green plaid bandana, dirt on cheek) staring coldly from the treehouse doorway; 10yo buzzcut boy (chipped tooth, striped cape) peeking over a bush on the far side of the wooden fence. Shallow DOF, subtle wind in leaves, building silence and pressure. 3-4.5s: [Push-in] Tree boy loads a stone into the leather pouch, draws the rubber bands hard, muscles tense, eyes locked. Speed ramp into slow-mo as the bands stretch. 4.5-6s: [Whip pan + slow-mo] Slingshot snaps. Stone launches in a high dramatic arc over the fence, camera orbiting the projectile, sunlight flaring off it. 6-7.5s: [Low angle] Buzzcut boy dives sideways behind the bush. Stone smashes into dirt, explosion of dust and grass in frozen-time impact. Freeze-frame on the dust cloud, then smash-cut. 7.5-9s: [Handheld tracking] Buzzcut boy rolls out, grabs a rock, fires back with raw arm power. Camera races with the counter-shot. 9-10.5s: [High angle from tree] Tree boy leans out, cape flaring, dodges the incoming stone. Near-miss clips a wooden plank, splinters fly in slow-mo. 10.5-12s: [Dynamic crane] Tree boy swings onto the tire hanging from the branch, using momentum to reposition higher. Speed ramp: normal to hyper-slow as he hangs and aims again. 12-13.5s: [Extreme close-up] His finger releases. Second stone flies. Freeze-frame on the stretched rubber bands the instant they snap forward. 13.5-15s: [Tracking the stone] Projectile screams across the fence line in accelerated motion, camera locked on it, background blurring. 15-16.5s: Buzzcut boy vaults the fence in a desperate leap, cape whipping. He lands hard, rolls, comes up firing another rock. 16.5-18s: Chaos exchange. Rapid 1.5s cuts: stone impacts, wood chips flying, both boys ducking and reloading, leaves exploding, tire spinning wildly. 18-19.5s: [Hero low angle] Tree boy stands tall on the platform edge, silhouette against the sun, draws the final powerful shot. 19.5-21s: Full slow-mo release. The stone leaves the pouch, camera races alongside it in a long tracking shot as it curves higher than intended. 21-24s: The stone overshoots the fence completely. It sails toward a neighboring house window. Extreme slow-mo approach, glass surface reflecting the golden light. 24-27s: Impact. Window shatters in spectacular slow-motion — glass shards exploding outward in every direction, freeze-frame on the peak shatter, then resume with cascading fragments. 27-30s: Smash-cut to the neighbor (middle-aged man) filling the broken frame, face twisted in pure rage, veins popping, absolute fury. Handheld shake, extreme close-up, final freeze on his furious expression. Photorealistic, heavy film grain, ultra-detailed textures, perfect motion blur, coherent physics, stable characters, cinematic lighting, movie-level stability.

Made with SeeDance 2.5 Prompt: Main Subject: Young Korean woman, 24, naturally attractive, realistic skin, minimal makeup, long dark hair loosely down. Wearing a simple casual outfit. Preserve her exact identity, facial features, hairstyle, and appearance throughout. Location: Quiet Seoul apartment or residential rooftop during a warm late afternoon. Soft natural sunlight, gentle breeze, everyday surroundings. Style: Ultra-realistic early-2000s Sony MiniDV home video. Completely candid and unstaged, handheld shake, imperfect framing, autofocus hunting, exposure shifts, faded colors, soft contrast, DV compression, slight motion blur and microphone noise. No stabilization or modern cinematic movement. 00:00–00:04: She stands casually in front of the camera, trying to tie her long hair into a ponytail. A few strands keep slipping out. 00:04–00:08: She tries again, struggling with the hair tie. Her expression gradually becomes adorably irritated as the ponytail falls apart once more. 00:08–00:11: She sighs, slightly scrunches her face in frustration, then suddenly notices the camera watching her. 00:11–00:15: She looks directly into the lens, breaks into a genuine laugh at herself, and shakes her head. A few strands fall across her face as she gives up fixing them. Audio: Natural location sound only—soft wind, distant neighborhood sounds, subtle clothing and hair movement. No music or added sound effects. Goal: Cute, spontaneous and intimate, like a tiny everyday moment accidentally captured on an old home video. Her irritation should feel natural and playful, ending in a genuine laugh toward the camera.

Created with Seedance 2.5 🌊🚲 Summer Convenience Store Run → Seaside Snack Break Prompt 👇 30s ultra-realistic vertical iPhone vacation video in a sunny Japanese coastal town. FORMAT: 9:16 portrait only. Handheld iPhone footage filmed casually by a friend. CHARACTER: Use the provided reference image as the ONLY identity reference for the main woman. Keep her exact face, hairstyle, skin tone, facial proportions and body proportions consistent throughout. No beautification, redesign or identity drift. STYLE: Natural summer day trip, youthful, playful, spontaneous. Real reactions, imperfect framing, slight camera shake, autofocus breathing, exposure changes, smartphone HDR, natural motion blur. No cinematic look, no posing, no beauty filter. 0–5s: Friends ride bicycles through a quiet Japanese coastal street. The main woman looks back when called, laughs and almost misses a turn. 5–10s: Outside a small convenience store, she drinks a cold drink and laughs while opening a snack. Camera briefly moves too close, then pulls back. 10–15s: The group walks toward the sea with snacks. She points at something near the ocean. A friend briefly blocks the camera as a joke. 15–20s: At a seaside viewpoint, she sits on a low wall. A friend tosses a snack toward her; she dodges it and laughs. 20–25s: Walking toward the beach, she turns around and walks backward smiling at the camera, nearly bumps into a friend, then laughs. 25–30s: They reach the shoreline. She walks near the water, smiles naturally at something off-camera, and the video ends casually with bright ocean light and slight lens flare. ENVIRONMENT: Narrow Japanese streets, bicycles, convenience store, vending machines, local houses, utility poles, summer greenery, ocean. AUDIO: Natural iPhone audio: bikes, footsteps, wind, birds, traffic, ocean waves, friends chatting and laughing. AVOID: cinematic shots, slow motion, perfect composition, influencer posing, beauty filters, plastic skin, facial drift, subtitles, logos. Try Seedance 2.5 👉

Prompt: Create a 20-second photoreal cinematic video set aboard the RMS Titanic on the night of April 14, 1912, in the final moments before the ship strikes the iceberg. The concept is: The world 10 seconds before disaster. For most of the video, everything should feel elegant, calm, and ordinary. The audience knows what is coming, but the passengers do not. Build tension through small physical cues and subtle changes in atmosphere, not melodrama. The last few seconds must show the actual beginning of the disaster: the first sight of the iceberg and the first impact with the ship. 0–3 SEC — ELEGANT NORMALCY Interior, first-class dining saloon at night. Warm golden electric light. Well-dressed passengers sit at tables eating dessert, sipping drinks, and talking quietly. A steward clears plates from a nearby table. Silverware glints. Crystal glasses catch the light. Everything feels luxurious, calm, and completely routine. 3–6 SEC — SUBTLE STILLNESS Camera slowly moves through the dining room. A waiter pours champagne at one table. A couple smiles and speaks softly. There is a strange sense of stillness beneath the elegance — the ship feels smooth, steady, and quiet on the cold Atlantic. No one appears concerned. No one knows anything is wrong. 6–9 SEC — SMALL PHYSICAL CLUES A water glass on a table gives a very slight tremble. A spoon softly rattles against fine china. One passenger pauses for a moment and glances up, as if they felt something faint through the floor, but then returns to conversation. The ship still seems normal. No alarm. No panic. 9–12 SEC — CUT TO THE DECK / LOOKOUT TENSION Cut to the dark exterior forward deck and crow’s nest atmosphere. Freezing black night. The sea is unnaturally calm. The camera looks forward over the bow into darkness. A lookout strains his eyes into the night air. His breath is visible in the cold. The silence feels ominous. 12–15 SEC — THE ICEBERG APPEARS Straight ahead in the darkness, a huge pale shape slowly begins to emerge: the iceberg. At first it is vague and ghostly. Then it becomes unmistakable. The lookout reacts. A bell is rung urgently. The tension spikes instantly, but there is still only a second or two before impact. 15–17 SEC — URGENT RESPONSE The camera stays forward-facing as the Titanic continues moving toward the iceberg. Crew reaction begins. There is sudden urgency — shouted warning, fast movement, immediate recognition of danger. The iceberg now fills more of the frame, looming just off the ship’s path. This is the final suspended second before collision. 17–20 SEC — IMPACT BEGINS The ship begins to turn, but too late. The starboard side makes contact with the iceberg. Show the first scraping impact beginning along the hull: a deep grinding sound vibration through the ship ice and frost breaking along the metal passengers inside briefly jolting or glasses trembling harder Do not jump ahead to flooding or sinking. End at the moment the collision has clearly begun. VISUAL STYLE Photorealistic prestige historical filmmaking. Authentic 1912 Titanic details: correct first-class dining room design period clothing historically accurate table settings accurate deck and night-ocean atmosphere realistic cold Atlantic darkness correct scale and appearance of the iceberg Lighting inside should be warm and elegant. Lighting outside should be extremely dark, cold, and minimal, with the iceberg gradually becoming visible out of the blackness. CRITICAL RULES The first 12–15 seconds must feel mostly calm and ordinary. Do not show panic too early. Do not show the ship already damaged. Do not jump ahead to chaos, flooding, lifeboats, or the sinking. The suspense should come from the audience knowing what is about to happen while the people onscreen remain unaware. The last few seconds must show the actual beginning of the disaster: the iceberg is seen the warning is triggered the ship begins to strike it The exact progression should be: elegant dinner → tiny vibrations → cut to cold dark deck → iceberg appears → urgent warning → first scraping impact begins

90s high school montage. Seedance 2.5 in @TopviewAIhq Prompt: [Generation goal] Reference-to-video. One continuous 20-second handheld camcorder montage following a High school student named Kim through four locations, shot to look like 1998 consumer VHS home video. [Reference roles] @Image 1 defines Kim: East Asian woman, late teens, black shoulder-length waves, open red and white plaid flannel over a white ribbed tank, light-wash straight-leg jeans, white chunky sneakers, thin gold chain. Use her face, hair and wardrobe only. Do not inherit the gray studio backdrop, the four-panel layout, the corner inset, or the neutral expression. [Stage 1 — approximately 0-5s] The operator walks through a doorway into a school class room of seated students. Kim sits turned sideways in her chair mid-conversation with one classmate. She catches the lens, breaks off the sentence, opens into a wide open-mouth smile and waves with her whole hand. No backpack anywhere on her. End state: Kim still waving, the classmate turning toward the camera. [Stage 2 — approximately 5-10s] High school corridor. Lockers line both sides. Kim steps into the middle of the hall and drops into a four-count hip hop step: shoulder rolls, a quick two-step, one spin. She loses the beat coming out of the spin, laughs at herself, and keeps going anyway. Two students edge past along the wall. End state: Kim doubled over, hands on her knees, laughing. [Stage 3 — approximately 10-15s] School cafeteria. Kim at a long table with three friends, tray in front of her, black backpack on the floor beside her chair. She says something, the table laughs, she points at one friend. End state: Kim leaning back in her chair, grinning. [Stage 4 — approximately 15-20s] Outside, brick building of the high school behind her. Kim works a piece of gum and blows a pink bubble that grows past her chin. At 18s a friend's finger enters from frame left and pops it. Gum flattens across her lips, cheek and nose. Half a beat of stillness, then both fold into laughing. End state: Kim peeling gum off her cheek, still laughing. [Camera] Handheld throughout, operator walking, eye level to chest height. Each location change is a hard in-camera cut with one frame of motion blur and a tape glitch. [Visual style] 1998 consumer camcorder recorded to VHS. Soft focus, visible horizontal field lines, chroma bleed and smear on the red plaid, white dropout streaks, windows blown out by auto-exposure, heavy gain grain indoors, cool fluorescent inside and warm daylight outside. [Audio] (thin hip hop track from a boombox down the corridor, stage 2 only) <room chatter, footsteps on tile, tray clatter, wind on the mic outside> American English, casual and unrehearsed, spoken by Kim in stage 1: {Are you recording? Oh my god.} Camcorder mic character: mono, thin, slightly clipped. [Maintain consistency] One Kim throughout, same face, hair and outfit in all four locations. No backpack on her body in any stage; it appears only on the dining hall floor in stage 3. Gum appears only in stage 4. Flannel stays open. No wardrobe change, no on-screen text, no subtitles, no logos, no smartphones, no clean digital video look.

Made with seedance 2.5 Prompt: Use @ Image1 as the visual baseline. Maintain a locked, static camera, a completely blurred muddy trench background with drifting smoke haze, floating embers, and the faint orange flicker of distant fires, and desaturated, gritty cinematic war-film color grading with heavy film grain. Ensure the paws remain strictly feline with no human fingers, and the helmet and rifle maintain rigid structural consistency without melting or warping. Execute this exact 15-second looped sequence: Start with only the very top curve of the green military helmet visible above the dirt mound. The white cat slowly rises vertically into frame, resting the miniature rifle on the dirt and taking aim. The cat squints its left eye tightly, grips the rifle with its right feline paw, and fires. The miniature rifle kicks back with sharp, rapid recoil, accompanied by a quick muzzle flash and a puff of smoke. An enemy round strikes the top of the dirt mound immediately after, blowing a spray of dirt into the air. The cat drops straight down behind the mound until it is completely hidden. While hidden, the cat lifts the helmet off its own head with both feline paws and balances it upside-down on the muzzle of the rifle. The cat's head is now bare, ears flattened. The cat slowly raises the rifle, and the helmet rises above the dirt line alone, wobbling gently as bait. The cat itself stays fully hidden behind the mound. The decoy helmet holds still above the dirt line. A long, deliberate pause. A rifle crack rings out and the helmet is knocked violently sideways off the barrel with a bright spark and a metallic ring, tumbling out of frame. In the same instant the cat pops up bare-headed, ears flat back, fur ruffled, eyes narrowed, shoulders the rifle and fires straight ahead. Sharp recoil, muzzle flash, and smoke. The cat holds the aim through the drifting smoke, then sinks slowly straight back down behind the dirt mound, catches the falling helmet, and sets it back on its head, ending exactly as the video started. Audio: a continuous low rumble of distant artillery, muffled shell impacts thudding at irregular intervals, fires crackling off-screen, a shell whistling overhead, and wind moving across open mud. Sharp dry rifle cracks on each shot, a bright metallic ping and ricochet whine when the decoy helmet is hit, and dirt pattering down after each impact. No music and no voices.

Prompt: Rapid aerial sweep over a vast 3D map of a fantasy continent, built as an intricate clockwork model — aged parchment terrain, carved-wood forests, hammered-bronze mountains, hand-etched rivers. Warm amber and sepia palette, candlelit shadows, volumetric light rays. ONE continuous unbroken shot, fast momentum, no cuts. The camera follows a winding river as a leading line, banking hard through whip-turns, then dives low toward a coastal city that self-assembles as it approaches — docks unfold, towers telescope upward, flags snap open. Camera threads BETWEEN the towers and pulls up, using a foreground spire to wipe the frame, then cranes over a bronze mountain range where a fortress gate erupts from the rock. It swoops down THROUGH a spinning brass gear (occlusion transition into darkness and back to light), skims a bridge that extends across a moat toward a walled capital whose spires spiral up, then banks skyward and cranes back to reveal the full glowing map beneath a rotating golden astrolabe sun. Every location connects by motion — rivers, roads and gears act as transitions, camera speed constant and sweeping. SFX: deep gear rotations, ratcheting clicks, creaking wood and metal, whoosh of air on each pass, resonant brass chime on the final reveal. No music. Native 4K, ultra-detailed miniature craftsmanship, tactile realism, cinematic depth of field.

Created this cinematic fantasy fight scene using @lovart_ai + Seedance 2.5 . Prompt ⤵️ 30-Second Cinematic AI Video Prompt Create a highly realistic cinematic fantasy action-comedy scene inside an ancient wooden tavern at night. The environment is dark, atmospheric and richly detailed, with wooden walls, traditional furniture, hanging lanterns, warm orange candlelight, smoke and subtle background patrons. Character 1: A beautiful pale-skinned female warrior with long platinum-white hair, wearing a black-and-deep-red leather fantasy warrior outfit with metallic details. She carries an oversized sword and has a serious, confident expression. Character 2: A heavy-set East Asian woman wearing simple loose grey traditional clothing. She initially appears relaxed and harmless but gradually reveals surprising strength and a powerful personality. Scene 1 — Entrance | 0–5 seconds The tavern doors suddenly open. The white-haired warrior slowly walks inside carrying her huge sword. Low-angle cinematic shot following her boots, then tilt upward to reveal her face. She looks around the tavern and notices the heavy-set woman sitting at a table. Warrior, in natural English, confidently: “Are you the one they call the Dragon of this tavern?” The woman looks up from her food, completely unimpressed. Woman: “That depends. Who’s asking?” Use realistic lip-sync, natural facial expressions and subtle background tavern sounds. Scene 2 — Confrontation | 5–10 seconds The warrior approaches the table and places one hand on her sword. Warrior: “I came here to collect a debt.” The woman casually takes another bite of food. Woman: “Then you came to the wrong table.” The warrior becomes irritated. Warrior: “I don't think you understand who you're dealing with.” The woman slowly smiles. Woman: “Oh, I understand perfectly.” Scene 3 — Fight Begins | 10–17 seconds Suddenly the warrior swings her enormous sword. Use dynamic handheld cinematography and fast martial-arts choreography. The woman quickly moves aside, surprisingly agile despite her size. The sword crashes into the wooden table, sending pieces of wood and food flying. The woman grabs the warrior and throws her backward. Warrior, shocked: “Wait… you're actually strong?!” Woman: “You should have asked before starting a fight.” Scene 4 — Fire Attack | 17–23 seconds The warrior gets back up and charges again. The woman takes a deep breath. A mysterious orange glow begins building inside her mouth. The warrior stops suddenly. Warrior: “What the hell are you doing?” The woman exhales a powerful burst of magical fire toward her. Create a dramatic, realistic fire effect illuminating the entire tavern. The warrior narrowly avoids the flames but gets thrown backward by the force of the blast. Scene 5 — Defeat | 23–27 seconds The warrior crashes onto the wooden floor. The woman walks toward her calmly. The warrior tries to crawl away and looks completely stunned. Warrior: “Okay… I admit it. I underestimated you.” The woman steps closer and places one foot firmly beside the warrior, looking down at her. Scene 6 — Final Shot | 27–30 seconds Extreme close-up of the woman's face. She looks directly at the warrior with a calm, almost amused expression. Woman, quietly: “Next time… ask questions before swinging a sword.” Cut to black. Visual Style Ultra-realistic cinematic fantasy film, high-budget movie production, realistic human faces, detailed skin texture, natural hair movement, physically accurate clothing, realistic sword physics, believable martial-arts choreography, cinematic fire VFX, volumetric smoke, warm lantern lighting, dramatic shadows, shallow depth of field, realistic motion blur, atmospheric particles, detailed medieval tavern environment, professional Hollywood cinematography. Use a mixture of wide establishing shots, low-angle hero shots, medium dialogue shots, close-ups, over-the-shoulder shots and dynamic tracking shots during the fight.

I'm having way too much fun pushing Minimax H3 for cinematic typography. This time: an entire mountain becomes the title. Massive scale. Ancient stone. Pure destruction. PROMPT 👇 Ultra cinematic title sequence: A gigantic mountain cliff begins collapsing into a deep canyon. Entire rock faces fracture while dust clouds rise thousands of feet into the atmosphere. Giant stone monoliths rotate in slow motion through the air, colliding with breathtaking force. Rock fragments, dust, and debris gradually assemble the monumental word "DOMINION", carved from ancient stone scarred by centuries of erosion. The canyon echoes with a deafening rumble before the title crumbles into dust that is carried away by the wind. Monumental scale, ancient power, Hollywood epic title design.

Created with Seedance 2.5 | 30s | 16:9 Create a photorealistic smartphone travel vlog of one woman taking her first trip to Tokyo. Identity & continuity: Use @Image1 only for identity. Preserve her exact facial identity, skin tone, hairstyle, body proportions, outfit, handbag, suitcase, and overall appearance throughout. No changes or identity drift. Outfit: Cream-beige cropped trench coat, ivory knit top, dark charcoal wide-leg trousers, white sneakers, black crossbody bag, silver watch, natural makeup. Style: Authentic handheld smartphone footage—subtle shake, autofocus breathing, exposure shifts, natural motion blur, realistic colors. No commercial/cinematic look. Sequence: 0–4s: Airport arrival, pulling suitcase. Selfie: “Okay… this is actually happening.” 4–8s: Departure terminal; casually film architecture and flight screens. 8–12s: Airplane window, wing above clouds. Selfie: “I can't believe I'm going to Tokyo.” 12–16s: Landing/taxiing in Japan; airport activity outside. 16–20s: Exits airport with suitcase; train, vending machines, bicycles nearby. 20–24s: Quiet Tokyo neighborhood with cafés and storefronts. Selfie: “Everything looks so different.” 24–28s: Evening Tokyo street, crossing, lights and pedestrians; excited selfie. 28–30s: Close selfie while walking: “Alright… let's see where this takes me.” Reaches toward phone; recording ends. Audio: Only realistic diegetic sounds—airport ambience, suitcase wheels, footsteps, airplane sounds, train, traffic, conversations, wind and Tokyo street noise. No music. Negative: No subtitles, overlays, logos, watermark, face distortion, identity drift, outfit/hairstyle changes, duplicate person, CGI look, beauty filter, fantasy elements, or impossible locations. Overall: casual, spontaneous, warm, believable real-world journey from airport → flight → Japan → Tokyo → evening city.

GPT Image 2 and Seedance 2.5 Prompt: Style: Cinematic live-action seamlessly blended with 2D hand-drawn line-art animation, using a playful rotoscoping-inspired transformation effect. Visual Direction: First-person POV throughout. A hand holding a blue pencil appears at the bottom of the frame, seemingly tracing over real-world objects. Wherever the pencil moves, the live-action environment gradually transforms into expressive 2D illustrations. Scene Transitions & Subjects: - Train: A white-and-green train travels across an elevated urban overpass. As the blue pencil traces along it, the train smoothly converts into clean black-and-white sketch line art, while retaining its movement and perspective. - Cat: An orange tabby cat casually walks along a paved sidewalk. The pencil follows its movement, transforming the cat into a white hand-drawn cartoon with stylized illustrated shadows. - Bus: A light-colored compact bus drives through the street. As it passes beneath the pencil’s tracing path, it morphs into a colorful pastel illustration with loose, artistic hand-drawn outlines. - Girl: A woman sits naturally on a bench in a narrow urban alley, looking at her phone. The pencil traces around her and the surroundings, seamlessly turning the scene into a stylized 2D color-block illustration. Tone & Atmosphere: Bright natural daylight, contemporary urban environment, playful and whimsical mood, fluid camera movement, dynamic transformations, and seamless transitions between realistic footage and hand-drawn animation. Overall Effect: The video should feel as though the blue pencil is literally drawing reality into an animated world, with each subject transforming organically rather than appearing as a sudden cut or visual effect.

Seedance 2.5 is good at looking upstaged🤯 generated on arcads full prompt: Single unbroken handheld take throughout, 30 seconds total. Young Caucasian woman, long wavy platinum-blonde hair, blue eyes, wearing an oversized red and white vintage football jersey, black denim shorts, white crew socks, and white sneakers, sitting pensively on a sofa inside a large bright suburban Parisian house, summer daytime, sunlight through the windows, young adults dancing and laughing around her holding drinks, loud music implied by energetic crowd movement. 0-4s: She rises from the sofa empty-handed, walks toward a kitchen table lined with alcohol bottles and glasses. Handheld camera follows close behind her shoulder, slightly shaky. 4-8s: She pours herself a glass of champagne, sips it. Camera holds behind her, shaky, close distance. 8-13s: She weaves through the dancing, laughing crowd, some standing, some in groups laughing on sofas, toward a large glass bay window. Camera continues following close behind her, no cut. 13-16s: She steps outside into the garden, a big pool comes into view with young adults swimming and laughing inside it. Camera smoothly rotates from following behind her to facing her. She looks toward the pool with amusement, smiles, sets her champagne glass down on a small table beside a lounge chair. 16-20s: She removes her oversized football jersey, places it on the back of the lounge chair, then removes her black denim shorts, places them on the lounge chair, revealing a simple black swimsuit underneath. Camera holds a medium shot facing her, slight handheld sway. 20-24s: She walks toward the pool, camera repositions behind her as she walks, then she breaks into a run. Handheld camera follows close, shaky, matching her pace. 24-27s: She dives into the pool shouting “take care”, landing in the water among other laughing young adults. Camera dives in with her, briefly submerged underwater alongside the swimmers. 27-30s: She resurfaces laughing, joins the laughing adults nearby. Camera is up in the air, revealing a wide aerial view of the house, garden and pool. Bright natural summer daylight, warm sunlit color palette, candid documentary energy, slight handheld lens wobble throughout. Maintain her exact long wavy platinum-blonde hairstyle, blue eyes, oversized red and white vintage football jersey, black denim shorts, white crew socks, and white sneakers until she undresses, then maintain the same simple black swimsuit for the remainder. Background party guests and swimmers generic, unnamed. No text overlays, no flickering, no ghosting, no morphing artifacts, natural skin tones, stable continuous handheld motion, no hard cuts at any point.

Seedance 2.5 Prompt: 30 seconds | 16:9 | No music Ultra-photorealistic live-action, AAA Hollywood action style with premium VFX. One continuous dynamic shot, permanent 40° Dutch angle, low tracking, fast pursuit, whip-pans and quick orbiting. Real human performance, not CGI. IMAGE 1: Preserve exact identity, face, hair, clothing and proportions. Image1 has ONLY ONE ARM — RIGHT ARM ONLY. No left arm, hand, elbow, prosthetic or phantom limb. Sword is always controlled one-handed with the right hand. SETTING: Massive medieval royal plaza, castle walls, stormy sky, damaged market structures and dozens of armored soldiers surrounding Image1. 00–06s: Soldiers attack. Image1 narrowly dodges a sword, parries one-handed, vaults over an attacker and kicks another into a shield wall, then bursts forward with a black afterimage and blue-white lightning trail. 06–12s: Image1 runs across fallen shields, bends a spear with his foot and cuts it with a lightning-charged slash. He rolls, rises and immediately fights three soldiers using fast parries, kicks and a reverse sword strike. 12–18s: A dense spear formation attacks. Image1 slides underneath, cuts several spear shafts and spins into a powerful heel kick. Lightning jumps between broken weapons as soldiers are thrown backward. 18–24s: Surrounded by enemies, Image1 throws the sword upward, then uses rapid Taijutsu—low sweep, shoulder strike, high kick and jumping knee. Brief black afterimages emphasize his physical burst speed. No teleportation. 24–30s: Image1 catches the falling sword perfectly with his right hand. Electricity surges across the blade. A powerful one-handed slash and spinning kick create a compressed-air shockwave, blasting soldiers outward through dust and debris. Camera pulls back as Image1 keeps advancing and another wave of soldiers enters at 30 seconds. No ending pose. VFX: Restrained blue-white lightning, sparks, afterimages, shockwaves, dust and debris. Keep everything physically integrated with realistic lighting, shadows and reflections while Image1 remains fully authentic live-action. AUDIO: Wind, armor, sword impacts, electrical crackling, debris and battle cries. No dialogue or music. NEGATIVE: Left arm, two-handed grip, teleportation, clones, CGI human, digital double, cartoon look, gameplay camera, broken anatomy, face drift, costume drift, subtitles, logos, watermark.

AI just made this dessert look dangerously real 🤯 Created with Gemini Flash Prompt: A cinematic close-up food preparation video of Thai mango sticky rice and a refreshing pink lychee drink. A perfectly molded dome of sticky rice is plated beside thin fan-cut ripe mango slices, drizzled with creamy coconut sauce, topped with toasted sesame seeds and a fresh mint sprig. Ice cubes are added to a tall glass with a light pink drink, followed by a peeled lychee, a lemon wheel garnish, and fresh mint. The final shot showcases the elegant dessert and chilled drink on a rustic wooden table with a bright tropical countryside backdrop of lush green fields, palm trees, cozy huts, and distant mountains under a clear blue sky, featuring soft natural lighting, smooth camera movements, shallow depth of field, and ultra-realistic 4K cinematic quality.

Seedance 2.5 is absolutely insane it nails every region and culture on the planet. India's landmarks come to life so realistically, I feel like I'm actually traveling through Delhi🍛 👇Full Prompt INDIA TRAVEL VLOG — 30 SEC MUSIC CUT There are no reference images. Build the person from this description alone. SUBJECT: one Korean woman in her early 20s, the only person the video follows. Small oval face with a soft jawline, large round double-lidded eyes, a small straight nose, full coral-red lips, fair skin. The face established in the first shot is carried by every shot after it — same features, same colouring. She never turns into a different person and no look-alike appears. Her hair is black: from 0:00 to 0:15.5 it is in a wavy half-up ponytail tied high on the crown with a small pearl stud in each ear, and from 0:15.5 it is in one thick braid pulled forward over her left shoulder with a gold drop earring in each ear. Loose strands fall at the temples and nape throughout. Real Korean skin texture with visible pores — no doll-like AI smoothness, no beauty filter. The same face in every shot; she never turns into a different person. OUTFIT: from 0:00 to 0:15.5 she wears a cream ribbed knit short-sleeve top with a single row of pearl buttons down the front, a dark brown pleated wrap mini skirt, a tan leather belt, white socks to just below the knee, and dark brown loafer heels. From 0:15.5 to 0:30 she wears a cream sleeveless embroidered kurti, cream wide palazzo trousers, and a chiffon dupatta graduating from pale apricot into orange with a gold embroidered border, draped from her left shoulder across to her right hip with the end hanging to her calf. A thin gold bangle on one wrist, nude strappy sandals. The clothes and the hair change on the cut at 0:15.5; there is no changing shot and no shot of her redoing her hair. CAMERA: she is filming herself on her own phone. Most shots are selfies with her arm extended; objects and food are shot by turning the phone toward them. 16:9 landscape throughout, no black bars. 26mm phone lens, deep focus, neutral iPhone colour, handheld jolt on every step, autofocus landing a beat late, exposure lagging when she moves from bright to dark, a fingerprint smudge bleeding beside strong lights. No cinematic grading, no studio lighting, no glow fx. Background people stay out of focus and nobody looks at the camera for long. DIALOGUE: lines are short and land in one breath. An interjection plus one short sentence is fine, but never two complete sentences strung together. There is breath between them and they are never rushed into each other. Punctuation is used once — no doubled question marks or exclamation marks. The two Hindi greetings are not fluent — just phrases she has picked up. (0:00–0:02) — Guesthouse rooftop, Jaipur morning reveal. Selfie, low angle. Tilt up. She raises the phone from her face and tiers of pink sandstone buildings open up behind her. Washing lines and water tanks on the parapet, a single kite far off. The sky blows out white, then the exposure catches up. Bright grin. Dialogue: "Namaste~ 인도 왔다!" Transition: Hard cut on beat (0:02–0:04) — Old city junction. Selfie, handheld, stepping out between the vehicles. She swings the phone to her face while yellow-and-green auto rickshaws and motorbikes tangle past behind her — and in the middle of it a cow with painted horns stands completely still. Overcast daylight, deep focus. Dialogue: "아니 소가 왜 여기 있어ㅋㅋㅋ" Transition: Fast cut on beat (0:04–0:06) — Street chai stall. Selfie, arm short, static. She pushes an unglazed clay cup toward the lens until it fills half the frame, eyebrows raised, a thin skin set on the surface. An aluminium kettle hangs at the stall behind her, low wooden benches beside it. Transition: Snap cut (0:06–0:08) — Alley samosa cart. Insert, phone pointed down at her hands. The cart owner scoops samosas out of the black iron kadhai with a ladle, tips them onto a newspaper plate, spoons green chutney beside them, and lowers the plate into her two hands. Steam rises and oil spreads into the newsprint. Daylight under the cart's awning, a parked bicycle beside it. Transition: Cut on natural movement (0:08–0:09.75) — Hawa Mahal reveal. Tilt up. She raises the phone from her face until the pink sandstone facade fills the frame, its honeycomb of small windows continuing upward. Natural daylight, sky blown out over the top. Her mouth opens slightly. Transition: Soft cut mid-upward motion (0:09.75–0:12) — Panna Meena ka Kund stepwell. Selfie, arm extended high. Sitting at the rim she looks at the lens, then pushes the arm holding the phone out past the ledge and tips it downward. As the frame tilts down, the interlocking symmetrical staircases fold away below her without end, mossy stone and green water pooled at the bottom. A few pigeons lift off the wall. Dialogue: "이거 계단 몇 개야 대체" Transition: Cut as the frame reaches its lowest angle (0:12–0:13.75) — Jaipur bazaar. Follow shot from behind, phone low. She shoulders through the narrow passage with the samosa, finishes the last piece and folds the newspaper away. Stacks of fabric and heaps of glass bangles on both sides, people pushing past. Overcast afternoon light. Transition: Cut on forward movement (0:13.75–0:15.5) — Back seat of an auto rickshaw. Static, low. Her hands are empty now — the phone at knee height in one hand, the rail in the other. The street slides past the open side, the driver's back swaying in front. She looks out, not at the lens. The music drops one beat, then comes back. Transition: Snap cut (0:15.5–0:18) — Thali restaurant. From here the image is camcorder: blooming tape, light noise, late focus. The camera is fixed at eye level across the table, no phone or camera visible in frame. Medium. From here she is in the cream sleeveless embroidered kurti and the orange dupatta, her hair in one thick braid pulled forward over her left shoulder. It is the same face as in the first half. As she sits down on a wooden chair a waiter sets a steel tray in front of her — six small bowls of different curries and raita, rice in the middle, naan and papad and a wedge of lime around the edge. She holds up her right hand as if checking. Dialogue: "손으로 먹는 거 맞지?" Transition: Cut on the tray landing (0:18–0:20) — Tearing the naan. Macro, phone lowered to the tray, shallow focus. With her right hand only she tears the naan, dips it into a curry and lifts it, steam climbing. The steam crosses the lens and one side of the frame hazes over and clears. Transition: Cut with the naan still raised (0:20–0:22) — First mouthful. Selfie, handheld. She takes a big bite of the naan, closes her eyes and opens them. Her shoulders drop and she nods once at the lens. Late afternoon, the white fluorescent light of the restaurant. Dialogue: "미쳤다 진짜." Transition: Cut where her face softens (0:22–0:23.5) — Lassi. Selfie, close. She lifts a clay cup of lassi in both hands, takes a mouthful, sets it down and lets out a breath. A thin line of it is left on her upper lip and wiped away with the back of her hand. Transition: Cut back to phone image quality (0:23.5–0:27) — The Taj Mahal ★signature. She walks in beside the long watercourse on an extended-arm selfie and slowly lifts her head. Sunset light turns the white marble orange and the same colour sits inverted in the water beside her. Cypresses pass on both sides. The orange end of the dupatta lifts once in the wind, the same colour as the sunset sitting on the marble. Rack focus — her face sharp, the dome behind spreading. Practical light only, no glow. 60% slow motion. Transition: Soft cut with her head still up (0:27–0:30) — Ganges ghat, night reveal. From a close selfie she extends her arm all the way out until the wet steps and the river below come into frame. A small lamp on a leaf plate sits in front of her, and dozens of the same lamps drift scattered across the water. Wind pushes her bangs across her eyes once. Natural night phone exposure. Dialogue: "그럼 모두 alvida~!" She waves broadly at the camera and that same hand covers the lens as the frame goes dark. CONSISTENCY: the same face in every shot. The clothes and the hair change once, at 0:15.5, and are identical within each half. Hands are natural with no smeared fingers, and food is only ever picked up with the right hand. No text, no subtitles, no logo and no watermark on screen. No filming equipment is visible in frame. FINALLY: this has to look like footage a real person shot on a phone. Nothing about the face, the skin or the movement should read as AI-generated. Every line of dialogue is enunciated clearly, syllable by syllable — no slurring, no swallowed syllables, no clipped word endings, and the tense consonants and final consonants of Korean are pronounced accurately. Expressions and gestures are not performed; they are what the moment would naturally produce.

이 커플의 영상도 아래와 같은 간단한 프롬프트로 Seedance2.5에서 생성했습니다. https://x.com/seiiiiiiiiiiru/status/2088916415541825829?s=20 ---------- エモーショナルな思い出のような映画的な30秒の少年少女の会話シーン。 公園のブランコにそれぞれ座る中学２年生。 少女「ねぇ…？」 少年「…なに？」 少女「私たちって…いったい何なのかな？」 少年「なにって…そりゃ…」 少女「…？」 少年「AIだろ？」 少女「…うん。…だよね。」 ---------- 복잡한 구성으로 생성하는 경우에는 어쩔 수 없이 프롬프트가 길어지지만, 초보자분들은 간단한 프롬프트로 짧은 초수부터 시도해 보는 것을 추천합니다.

模型：Omni flash 提示词：10‑second SNS vertical short‑video, aspect ratio 3:4, 30fps. Setting: Beijing summer street, August midday 13:00, hot humid sunny weather 35℃, strong harsh sunlight, distant developing grey‑white cumulus clouds hinting afternoon thunderstorm, dry ground. First‑person smartphone POV shot, left‑hand holding phone out‑of‑frame, live‑action phone footage composited with rough cute 2D hand‑drawn frame‑by‑frame animation. Ordinary real‑world Beijing street scenes, no famous landmarks, no readable signs or text. Real human right hand enters frame from bottom‑right corner, holding fully fictional childish cardboard toy water‑gun‑style graffiti morph‑device; short chubby asymmetrical white cardboard bulk, cyan‑blue circular scanning ring on front, short clumsy toy‑water‑gun‑like grip, loose uneven black marker outline, visible white paper fiber, sparse cyan‑blue & light‑pink colored‑pencil hatching, single circular energy slot on side, no gun‑barrel, no weapon‑like structures. Maintain identical shape, color, size and hand‑grip posture for right hand and graffiti morph‑device across all shots. Press circular trigger button, scanning ring emits small cyan‑blue hand‑drawn ripple plus few black ink dots only, no shooting, no explosion, no flame, no damage. Ripple hits target, 0.6‑0.8‑second hand‑drawn lock‑on frame appears: wobbly black loose lines, cyan circles, light‑pink scale ticks, jittery graffiti markers tightly sticking onto target respecting perspective and occlusion. After lock‑on finishes, only the targeted everyday object transforms in‑place into rough cute 2D hand‑drawn animation; real‑life humans, streets, buildings, trees and other objects remain live‑action smartphone footage. Four shots connected by clean hard‑cuts only, no fades, no transition effects. Each shot fully shows complete sequence: real‑world target → lock‑on effect → in‑place transform to hand‑drawn version before cutting. [0.0‑2.5s Shot1: takeout thermal box beside subway entrance]: Ordinary Beijing subway entrance exterior, glass canopy, light‑grey stone ground, yellow tactile paving, street trees, no readable station name. Strong midday sun creates bright specular reflection on glass and pavement. Parked electric bicycle at shadow edge, rear rack mounts plain text‑free light‑grey square takeout thermal box. Slight forward walking camera motion. Right hand with graffiti morph‑device enters bottom‑right, aims at thermal box. Trigger pressed, cyan‑blue scanning ripple hits box, lock‑on frame adheres accurately respecting perspective and bicycle occlusion. After ~0.7s, only thermal box transforms in‑place into 2D hand‑drawn version: white‑paper box body, boiling jittery black outline strokes, cyan‑blue edge trim, light‑pink buckles, sparse ink specks; mounting position, dimension, orientation and shadow direction stay exactly unchanged. Electric bike, glass canopy, paving, passers‑by stay live‑action. [2.5‑5.0s Shot2: folding paper fan at grey‑brick hutong entrance]: Beijing grey‑brick hutong entrance under dense locust‑tree shade, dappled sunlight filtering through leaves, background air‑conditioning outdoor units, stone steps, slowly passing pedestrians, no readable house number. Passer‑by in light summer clothes walks left‑to‑right, only lower‑shoulder‑to‑back visible, naturally waving light‑wood folding fan. Graffiti morph‑device tracks fan from bottom‑right. Trigger pressed, lock‑on frame follows fan swinging angle and hand occlusion. After ~0.7s, only folding fan becomes hand‑drawn 2D animation: white‑paper fan surface, loose black‑line fan ribs, cyan‑blue and light‑pink colored‑pencil diagonal hatching. Real human hand, clothes, body and hutong environment keep live‑action. Hand‑drawn fan keeps natural swinging rhythm driven by real human wrist, never detaches from palm. [5.0‑7.5s Shot3: wide‑brim sun hat outside commercial street crossing]: Pedestrian waiting zone outside modern Beijing commercial street, light‑colored pavement, metal bollards, tree pits, blurred shared‑bikes in background, scorching bright sunlight, humid hot air. Passer‑by wearing plain light‑beige wide‑brim sun hat walks slowly from front‑right toward front‑left, back view only, no clear facial features. Slight laggy tracking camera movement. Graffiti morph‑device aims at hat, cyan‑blue ripple travels short distance, lock‑on frame strictly follows hat brim elliptical perspective, head occlusion and walking bounce. After ~0.6s, only sun hat transforms into white‑paper textured hand‑drawn hat: irregular black brim outline, cyan‑blue hatching on hatband, light‑pink marker patch on hat top. Passer‑by hair, body, clothes and whole street stay live‑action. Hand‑drawn hat bounces subtly with walking pace, firmly sits on real human head. [7.5‑10.0s Shot4: cold‑water sports bottle on shaded stone pier]: Street‑corner tree shade, light‑grey stone pier holds text‑free translucent cold‑water sports bottle with carry‑loop, real condensation water droplets on bottle surface. Background sun‑blasted Beijing street, distant slow‑walking pedestrians, rounded summer cumulus clouds in sky. Slight camera push‑in. Graffiti morph‑device from bottom‑right aims at bottle. Trigger pressed, lock‑on frame orbits around bottle for ~0.8s. Only sports bottle transforms in‑place into white‑background black‑line 2D hand‑drawn bottle: translucent white‑paper fill, cyan‑blue bottle cap, light‑pink carry‑loop, colored‑pencil hatching, tiny ink specks. Bottle‑to‑stone‑pier contact point, gravity, perspective and original shadow direction fully preserved. Several wiggly cyan‑blue hand‑drawn cold‑steam lines rise from bottle outer surface, tiny black ink specks float upward like small bubbles. Video ends on frame with swaying cold‑steam lines and slowly drifting distant clouds. 2D hand‑drawn animation style: uneven wobbly black outlines, white‑paper fill, colored‑pencil & marker hatching texture, ink grain, slightly over‑bleeding color blocks, frame‑by‑frame subtle boiling‑line jitter on every drawing frame, child‑like rough cute hand‑drawn aesthetic; no smooth vector graphics, no 3D CG render, no plastic material look, no neon sci‑fi hologram glow. After transformation, converted object strictly inherits original pixel position, scale, orientation, motion trajectory, gravity weight, perspective, occlusion relation, contact anchor points and shadow direction. Only target object visual material gets replaced; background, real‑human figures and supporting objects never animate. Lock‑on frame must stay tightly aligned to target, no misalignment; graffiti morph‑device never changes shape or color across hard‑cuts. Camera spec: smartphone main‑camera 1x wide‑angle, equivalent focal length 24‑28mm, NO 0.5x ultra‑wide / fisheye lens. Electronic image stabilization enabled, preserve subtle natural hand‑held micro‑shake, minor vertical bounce during walking, slight tracking lag, tiny auto‑exposure fluctuation. Right hand and morph‑device near‑lens appear slightly larger with soft minor softness; street background and target objects stay reasonably sharp. No heavy shallow‑depth‑field bokeh blur. Color grading: bright clean pastel cyan‑pink pop aesthetic, keep Beijing street photoreal believability. Sky & glass highlight tint soft pale cyan‑blue, grey brick & stone keep neutral clean tone, skin & warm surfaces tint soft peach‑pink, foliage low‑saturation olive‑green & teal‑green. Exposure lifted +0.5 stop, elevated black‑and‑white levels, low‑to‑medium contrast, subtle smartphone HDR, gentle digital sharpening, light SNS social‑media filter texture. No heavy film grain, no strong lens flare, no teal‑orange cinematic grading, no crushed deep black shadows. Audio: zero dialogue, zero voiceover, zero background music. Only soft button‑click SFX, gentle electronic lock‑on beep sound, marker‑pen‑on‑paper scratching sound, faint ink‑dot pop sound, mixed with quiet real‑world Beijing summer street ambient noise; sound effects precisely sync with four lock‑on trigger moments. Strict prohibitions: no on‑screen text, no titles, no logos, no watermarks, no brand / shop / station / license‑plate readable text. No firearm‑shaped props, no muzzle‑like structures, no bullets, no shooting impact, no attack, no explosion, no fire, no physical destruction. Never animate full‑frame image, never animate real human bodies. No target object drifting, geometry warping, proportion mutation, clipping‑through‑geometry, detaching from contact‑surfaces. Graffiti morph‑device must not alter shape or color. No deformed hands, extra fingers, missing fingers, left‑right‑hand swap errors. No extreme fisheye distortion, no heavy background bokeh, no excessive glowing bloom, no violent camera rotation, no slow‑motion, no black‑out frames, no freeze‑frame.

Seedance 2.5 1080p on @TapNow_AI Prompt: 7-shot handheld phone vlog montage, photorealistic snapshot realism, varied angles/cuts, slight tilt/shake, harsh fluorescent gym light mixed with warm high-window afternoon sun, fine film grain. Woman Image1: use reference only for face/hairstyle; fitted heather-grey short-sleeve athletic top, black leggings, tied-back hair with loose strands. Alone in a quiet scuffed gym corner with folded mats, spotted mirror, bench with rolled yoga mat/water bottle. Natural Korean dialogue/reactions, live ambient SFX only, no music/text/logos/watermarks. 1 (0-2s): Rope catches her foot; she stumbles and laughs. Close, low, drifting camera. “아 뭐야 벌써”. 2 (2-4s): Side close-up of hands/rope resetting; wrist flicks, rope slaps floor twice, then finds rhythm. Shaky follow. 3 (4-6s): Straight-on steady jump-rope rhythm, focused half-smile, quicker breathing, rope blur. “하나, 둘, 셋…” 4 (6-8s): Rope hits shin; she winces/laughs, hops once, restarts. Slight upward angle, slow push-in. “Okay, still counts.” 5 (8-10s): Faster set, sweat at hairline, heavy breathing, rapid rope beats. Straight-on, gentle sway, motion blur. 6 (10-13s): Rope tangles; she stops, doubles over laughing, hands on knees, shoulders shaking. Camera tilts down. “아 못하겠다 진짜”. 7 (13-15s): Still laughing/breathless, slowly straightens, wipes eyes, drops rope, grinning. Camera drifts back/up and lingers. Only her throughout; authentic fumbling, laughter, breathing, footsteps, rope impacts/drop, faint distant gym ambience. Do not depict/copy the reference image itself.

LIVE-ACTION CINEMATIC REALISM. Every frame is photographed as real live-action footage: a real adult woman, real city, real pedestrians, real domestic pets, and a real enormous mascot plush statue. Real skin with pores, real wavy blonde hair with individual strands, real cotton, nylon, leather, metal, wet asphalt, glass, concrete and plush fibers. No illustration, painting, anime, CGI, 3D render, or artificial surface. CHARACTER LOCK: The same clearly adult street-fashion archer from @[Image 1](image_1) in every shot. Preserve her exact face, natural facial geometry, dark eyes, skin tone, expression, extremely long straight wavy blonde hair, thick blunt bangs, and royal-blue backward baseball cap with small decorative details but no readable text. She is 178 cm tall, XS size, very slim elongated runway-model proportions. Her outfit never changes: oversized white-to-light-grey graphic cotton T-shirt with a deep plunging neckline that that accentuates her large cleavage., faint doodle motifs but no readable lettering, dusty rose-pink wide cargo trousers with dimensional pockets and heavy folds, black belt with silver hardware, colorful keychain charms, layered bead necklaces, and red low-top sneakers with white soles. ARCHERY GEAR: She carries a long dark laminated-wood recurve bow with a leather-wrapped grip, visible taut string, realistic curved limbs and subtle wear. A dark-brown leather quiver crosses diagonally over her back, holding six to eight arrows with natural wood shafts and cream-white feather fletching. She wears matching brown leather archery gloves with stitched palms, reinforced knuckles and wrist straps, plus a fitted leather forearm bracer on the bow arm and a slim finger tab or draw glove. The archery equipment is integrated into her street-fashion styling, not a historical costume. LOCATION: A dense urban shopping district and pedestrian plaza at blue hour, wet asphalt reflecting storefront light, concrete towers, glass façades, crosswalk signals, street lamps, overhead transit lines, narrow side streets, urban haze, and a large original unbranded mascot plush statue installed in the center of the plaza. The mascot is enormous, approximately three stories tall, with a soft cream-and-pastel plush body, oversized rounded head, expressive embroidered eyes, visible seams, padded limbs and tactile fabric. No copyrighted character, logo or readable text. STORY — 12 SHOTS, ONE CONTINUOUS EVENT: 1. 0–1.5s, LOW WIDE. Camera whips right past two foreground pedestrians and a street barrier. She bursts through the wet plaza at a low run, bow held low, red sneakers splashing lightly, dusty rose cargo fabric and extremely long hair streaming behind her. 2. 1.5–3s, EXTREME CLOSE, LOCKED. A commuter’s eyes appear above a scarf behind a transparent umbrella. The eyes snap toward the plaza as nearby city lights reflect across the wet plastic. 3. 3–4.5s, CLOSE, FAST PUSH-IN. Two pedestrians and their leashed pets in the distance suddenly turn toward the giant mascot statue. A small dog raises its ears while a cat inside a carrier shifts and looks outward. 4. 4.5–6s, TRACKING MEDIUM. Camera runs beside her through the moving crowd at shoulder height. People step aside, coats and shopping bags sway, her long black hair, oversized T-shirt and cargo trousers carry real momentum as she reaches toward the quiver. 5. 6–7.5s, LOW ANGLE CLOSE. Her red sneaker plants on the wet pavement. She straightens hard and raises the dark recurve bow into frame. The giant mascot statue is visible far behind her between buildings. 6. 7.5–9s, MEDIUM FROM HER LEFT. Camera orbits behind her as she draws an arrow from the quiver, nocks it and pulls the string to full draw. The bow limbs flex visibly; pedestrians freeze at a safe distance, watching. 7. 9–10.5s, MACRO, LOCKED, razor-shallow focus. Her leather-gloved fingers grip the string, stitched glove texture, finger tab and arrow shaft sharply resolved. Her eye, blue cap and city lights remain soft behind. The string creeps back the last centimetre. The crowd noise drops for one hard beat. 8. 10.5–12s, WIDE FROM BEHIND AND ABOVE THE ARROW, LOCKED. She releases. The string snaps and blurs. The arrow tears forward between streetlights and glass buildings toward the giant mascot statue. 9. 12–14s, MEDIUM CLOSE TRACKING THE ARROW. The arrow breaks apart in mid-air: its front half dissolves into millions of bright rose-pink glitter particles streaming behind it, each particle catching the city light separately while the feathered tail remains intact. The arrow has no point. 10. 14–16s, WIDE LOW ANGLE. The glitter stream strikes the plush mascot’s chest and races through its seams as a bright rose-pink wave. Plush fibers ripple outward from the impact point. Pink reflections pulse across the wet pavement and nearby windows. 11. 16–18s, WIDE LOW ANGLE, FAST PULLBACK. The enormous mascot plush statue activates and rises into a fully living presence. Its padded head lifts, embroidered eyes illuminate softly, oversized arms move, seams flex, and rose-pink glowing fabric appliqués spread across its plush surface from the impact point. It remains clearly soft, textile-based and physically heavy, not a CGI monster. 12. 18–20s, FINALE, WIDE FROM BEHIND. Camera rises behind her as she lowers the bow and lets her weight settle naturally onto one hip. She looks up at the awakened mascot with a small confident smile. Pedestrians gather and react, some stepping backward, others pointing or raising phones. Dogs tug lightly at their leashes, a small pet in a carrier shifts, and the giant mascot waves slowly above the plaza. Glitter continues to fall through the city light. Vary the framing aggressively: extreme close-up, macro, close, medium, tracking, low wide and overhead. Never repeat the same shot type twice consecutively. Shots 2, 7 and 8 are locked; all others use clear camera movement. Every action must be instantly readable. LOOK: Photorealistic live-action large-format colour film with a refined three-strip Technicolor response. Concrete, glass, asphalt and distant architecture remain cool blue-grey and slightly desaturated. Saturated colours are the royal-blue cap, dusty rose cargo trousers, red sneakers, colorful keychain charms, rose-pink glitter, and the cream-and-pastel mascot plush. Fine photochemical grain, subtle halation on streetlights and glitter, realistic wet reflections, tactile cotton, nylon, leather, metal and plush fibers. Photographed reality, never drawn. LIGHTING: Blue-hour city light mixed with warm street lamps and storefront spill at approximately 4200K. Cooler sky fill at 4:1. Streetlights rim her long black hair, cap, bow and quiver. Thin urban mist and steam rise from vents and wet pavement. The glitter emits separate rose-pink specular points and casts soft pink reflections onto the mascot, asphalt, glass and her face. CAUSE AND EFFECT: Pedestrians and pets notice the strange event before the transformation, so their final reactions read as consequences. The glitter must visibly travel in sequence: arrow releases, arrow breaks apart, particles stream through the city, glitter strikes the mascot chest, light travels through the seams, the plush statue activates, then the crowd and pets react. Never cut directly to an already-animated mascot. PHYSICS: Real mass, real air and believable momentum. Bow limbs flex, string snaps, arrow flight has directional force, glitter behaves as millions of separate solid particles carried by fast air, and plush fabric compresses, ripples and rebounds with weight. Cargo fabric, T-shirt, hair, quiver strap and keychains respond to acceleration. Wet pavement reflects movement without becoming a mirror. The giant mascot moves slowly because of its padded mass. No slow motion; all shots run at real-time speed. AUDIO: Wet footsteps, light splashes, distant traffic, pedestrian chatter, shopping bags and coats moving, dog tags and leash tension, a faint pet sound, bowstring creak, sharp release snap, arrow hiss, crystalline glitter shimmer, plush fabric shifting, soft mechanical-like seam tension, crowd gasps and footsteps. No music. ON-SCREEN TEXT: None. No readable text, subtitles, numbers, captions, UI, logos, brand marks or watermark anywhere. OUTPUT: 20 seconds, horizontal 16:9, 4K, continuous cinematic live-action realism.

The Elven Princess Greets the Orcs Seedance 2.5 on @astorie_ai 720p, 30-second text to video generation. You can check the prompt below.

Made with Seedance on Runway. Prompt: 👇 CAMERA: Handheld vintage consumer camcorder footage. First-person footage recorded entirely by RILEY as she documents an afternoon working on an old sports car inside a small independent garage. The camera remains physically handheld