GPT Image 2.5 prompt
Create an experimental streetwear campaign poster viewed through a fictional thermal-imaging camera, vertical 4:5 aspect ratio. Four young adult models pose on uneven stone steps inside a dark rocky passage. One stands in the left foreground with hands near his pockets; another crouches behind him at center-left; a third stands upright in the middle background; the fourth stands closest to the camera on the right, with one hand behind his head. All wear oversized graphic T-shirts, loose baggy trousers with outlined star patches, and chunky sneakers.
Render the entire scene in a striking thermal palette: cold surroundings in near-black, midnight blue, and deep indigo; bodies and clothing glowing red-orange, bright yellow, and near-white at the hottest points. Preserve recognizable faces, fabric folds, shirt graphics, and rough rock textures beneath the thermal treatment. Subtle sensor noise, luminous edges, and a gritty fashion editorial atmosphere.
Overlay a precise white camera interface using small monospaced typography, thin tracking brackets around each face, corner framing marks, a central crosshair, a compass scale along the bottom, and a vertical temperature legend at the lower-right. Keep the interface organized and readable without covering the models.
Use these exact text elements:
Upper-left brand label:
“BMX THERMAL”
Technical information beneath it:
“SCAN: 004”
“MODE: HEAT VISION”
“SOURCE: HUMAN”
“SIGNAL: LIVE”
Top-center boxed label:
“CREATIVE ENERGY”
“DETECTED”
Upper-right:
“REC ●”
“SESSION 01”
“AFTER DARK”
Small readings beside the four faces:
“36.7°C”
“36.4°C”
“36.9°C”
“36.5°C”
Lower-left:
“ORIGINALITY: HIGH”
“FILTER: NONE”
“VISION: UNLIMITED”
Bottom-center campaign line:
“YOU CAN FEEL THE DIFFERENCE.”
Temperature legend label:
“ENERGY LEVEL”
Maintain the appearance of a captured thermal-camera frame, with strong depth, distinct silhouettes, and believable infrared color transitions. No natural skin colors, original reference branding, additional people, or watermarks.
GPT Image 2.5.
Prompt:
Create a sophisticated contemporary travel-art poster from the uploaded londscape photograph.
Strict 3:4 vertical format.
The upper half must remain the original landscape photogroph. Preserve all important environmental structures, natural textures, lighting, shadows, horizon, architecture, people and perspective. Do not regenerate the scene.
The lower half reconstructs the landscape's strongest narrative element using delicate photographic fragments, halftone printing, subtle paper fibers and fading edges.
Use a single horizontal geometric color field sampled from the landscape. Let the reconstructed subject or horizon physically intersect this field.
Keep the background most white with muted tones derived from the original image. Avoid saturated colors.
The result should resemble a limited-edition museum travel poster rather than a conventional photo collage.
MiniMax H3 Tokyo Kawaii Selfie | Fisheye POV + Neon Doodle Pop 💗
One camera. One girl. 15 seconds of pure kawaii energy. ✨
Try it 👉https://t.co/KzbFn4XfoC
Prompt 👇
16:9Create a 15-second high-energy cute Japanese GAL / gyaru music video, 16:9 horizontal.
Use the reference image as the ONLY character reference.
Keep exactly the same girl throughout the entire video:
same face, facial features, hairstyle, hair color, makeup, skin tone, body proportions, outfit, accessories, and long decorated pink nails.
ONE GIRL ONLY.
NO character switching.
NO outfit switching.
## MUSIC
Generate an original upbeat English pop song specifically matched to the visuals.
Music style:
cute energetic Y2K bubblegum pop + kawaii hyperpop + light dance-pop.
Tempo around 145–155 BPM.
The song should feel:
playful, sweet, youthful, catchy, energetic, sparkling and slightly chaotic.
Use:
punchy pop drums,
bouncy electronic bass,
bright synths,
cute digital chimes,
claps,
sparkling electronic accents,
short rhythmic transitions.
Female English vocal.
The vocal should sound youthful, cute, clear and melodic.
NO screaming.
NO random “ahhh”, “woo”, “bong”, “boom” or meaningless vocal sounds.
NO spoken dialogue.
Use real short English lyrics with a catchy hook, for example:
“Look at me, I’m shining bright,
Pink little stars all through the night,
Heart goes click when you come my way,
Sweet little chaos, make my day.”
The lyrics may be adjusted naturally to fit the melody and rhythm.
The girl MUST visibly open her mouth and accurately lip-sync to the English lyrics throughout the MV.
Her mouth shapes must clearly correspond to the sung words.
Do NOT keep her mouth closed while vocals are playing.
Facial expressions, mouth movement, hand gestures, camera movement and graphic effects must all synchronize with the musical beat.
## IMPORTANT MOVEMENT RULE
Every shot must use a DIFFERENT hand gesture, facial expression and body movement.
DO NOT repeatedly use peace signs.
DO NOT repeatedly push both hands toward the camera.
DO NOT repeat the same cheek pose.
DO NOT loop any previous movement.
Each new musical phrase introduces a visually different pose.
The performance should constantly evolve.
## CAMERA
Extreme ultra-wide-angle / fisheye action-camera look, approximately 12–14mm equivalent.
The camera stays close to the girl.
Strong playful perspective distortion.
When a hand briefly approaches the lens, it becomes dramatically larger in the foreground while her face remains visible.
Energetic handheld MV feeling.
Use different camera behavior for different shots:
micro push-in,
small side orbit,
quick tilt,
tiny pull-back,
low-angle close-up,
slight camera rotation.
Avoid repeating identical camera movements.
## LOCATION
Bright sunny Japanese city intersection,
Shibuya-inspired urban street,
blue sky,
colorful Japanese commercial buildings,
real pedestrians far in the background,
realistic outdoor daylight.
Keep the same general location throughout so the video remains visually coherent.
---
## SHOT 1 — 0–1.7s
Extreme fisheye close-up.
The girl looks directly into the lens while singing the first English lyric.
She makes ONE peace sign beside one eye while the other hand lightly holds her loose red tie.
She gives a cute pout, then instantly raises her eyebrows on the beat.
Small camera push-in.
Tiny pink sparkles pop beside her eye.
---
## SHOT 2 — 1.7–3.3s
Completely different action.
She drops the peace sign.
She points one index finger directly toward the camera while slightly leaning her shoulder forward.
The fingertip becomes oversized through fisheye perspective.
She continues clearly lip-syncing.
A glowing neon-pink heart traces a circle around her fingertip.
Camera slightly tilts sideways.
---
## SHOT 3 — 3.3–5s
New action.
She pulls her hand away from the lens and places both open palms underneath her chin in a cute flower pose.
She tilts her head strongly to one side.
Expression changes:
wide eyes → quick wink → bright smile.
Her mouth keeps moving naturally with every English lyric.
Tiny butterflies appear around her hair.
---
## SHOT 4 — 5–6.7s
New action.
She turns her upper body sideways and looks back toward the camera over her shoulder.
One hand flicks her pink-tipped hair.
The other hand stays naturally lowered.
She sings directly toward the lens with a playful confident smile.
Camera performs a quick small side orbit.
Pink glitter trails briefly follow her hair movement.
---
## SHOT 5 — 6.7–8.3s
New action.
She faces forward again and performs two cute alternating cat-paw gestures near her shoulders.
Left paw → right paw → both briefly raised.
She scrunches her nose and closes one eye.
Do NOT use peace signs in this shot.
Hand-drawn pink cat ears briefly appear above her head.
Tiny stars bounce around her fingers on the beat.
---
## SHOT 6 — 8.3–10s
New action.
She makes ONE small finger heart with one hand beside her cheek.
Her other hand points toward herself.
She mouths the strongest lyric of the hook with a large, clearly visible singing mouth shape.
She gives a cheeky side-eye, then looks directly into the lens.
Camera performs a subtle rapid zoom-out.
A neon heart pulses once behind her.
---
## SHOT 7 — 10–11.7s
New action.
She lifts both hands above her head and forms a large heart silhouette using both arms.
Her upper body sways once from left to right.
She smiles brightly while continuing accurate English lip-sync.
Camera moves slightly lower for a playful low-angle fisheye view.
Several translucent pink hearts float upward.
---
## SHOT 8 — 11.7–13.3s
New action.
She suddenly leans very close to the camera.
One hand gently cups the side of her mouth as if singing a secret directly toward the viewer.
The other hand stays behind her back.
She gives an exaggerated mischievous expression while clearly mouthing the lyric.
Camera makes a tiny clockwise rotation.
Pink handwritten doodles rapidly appear around her face.
---
## SHOT 9 — 13.3–15s
Final climax.
Use a completely NEW final gesture.
She steps back slightly, raises both open hands beside her head with fingers spread wide, then quickly brings ONE hand forward to make a tiny finger heart toward the lens.
She finishes the final English lyric with a huge joyful smile and a quick wink.
On the final beat:
pink hearts,
butterflies,
stars,
sparkles,
glitter,
and cute hand-drawn GAL doodles
burst outward around her.
End on her smiling directly into the lens.
---
## EFFECT STYLE
Effects must react to specific physical movements and musical beats.
Use:
hot-pink translucent hearts,
neon butterflies,
sparkling stars,
hand-drawn Y2K doodles,
glowing fingertip trails,
tiny glitter particles,
cute animated cat ears,
small handwritten pink symbols.
Do NOT keep effects constantly on screen.
Effects should appear briefly on strong beats, then disappear.
Effects must follow the girl's hands, fingertips, hair movement and facial expressions.
They should feel like playful Japanese GAL sticker graphics integrated into live-action footage.
NOT fantasy magic.
NOT cyberpunk.
NOT heavy VFX.
## VISUAL STYLE
Realistic live-action footage.
Cute Japanese GAL / gyaru aesthetic.
Y2K Japanese pop MV.
Bright summer daylight.
Realistic skin texture.
Glossy pink nails.
Energetic social-media music-video aesthetic.
Strong fisheye perspective.
Fast but readable visual rhythm.
Cute, colorful, playful and slightly chaotic.
## RHYTHM
Every 1–2 seconds introduce something visually new.
Gesture changes must land on beats.
Expression changes must land on beats.
Graphic effects must land on beats.
Camera accents must land on beats.
The girl's lip movement must continuously follow the English singing.
The sequence should feel intentionally choreographed to the music rather than random posing.
## NEGATIVE REQUIREMENTS
No repeated poses.
No repeated peace signs.
No repeated double-hand lens pushes.
No looping gestures.
No frozen mouth during vocals.
No random talking.
No screaming.
No meaningless vocal sounds.
No subtitles.
No lyrics displayed on screen.
No text overlays.
No scene duplication.
No duplicate hands.
No extra arms.
No extra fingers.
No fused fingers.
No warped face.
No sudden character changes.
No outfit changes.
No excessive motion blur.
No slow cinematic pacing.
#MiniMaxH3 #MiniMaxDesign
Create with GPT IMAGE 2
Prompt:
Use the person in the reference image as the main character, suitable for both male and female subjects. Preserve the exact facial identity, facial structure, age, skin tone, and natural expression. Do not change the person’s identity or create an artificial/plastic-looking face.
Create a 4:5 vertical street-editorial photograph set in a stylish urban café alley with a large black brick wall. On the wall, create a huge black-and-white realistic mural depicting the same person from the reference image, holding a metallic coffee pot and pouring coffee downward.
The real person stands directly below the mural, holding a coffee cup positioned perfectly under the stream of coffee coming from the mural, creating a fun, artistic, interactive, and viral visual effect.
For a female subject:
Wear an oversized white shirt with a relaxed, slightly loose fit, youthful and modern styling, naturally tucked in, paired with jeans or dark-colored trousers and clean sneakers. Avoid a stiff corporate look or tight-fitting clothing.
For a male subject:
Use a modern urban-casual outfit that looks youthful, clean, stylish, and contemporary.
The mural should have a realistic street-art aesthetic, featuring black and white tones with bold orange accents, expressive brush strokes, paint drips, and short graffiti typography such as “BỪNG NĂNG LƯỢNG” or “CÀ PHÊ MỖI NGÀY.”
Use neutral daylight with natural skin tones and no excessive yellow color cast. Camera angle should be slightly low and editorial, showing the subject full-body with a relaxed, natural pose. Do not use a chin-on-hand pose.
Ultra-photorealistic, cinematic depth of field, realistic hands, accurate coffee stream, natural body proportions, authentic skin texture, highly detailed, 8K quality.
Do not copy the exact composition of the reference/sample image. Keep only the core concept of a real person interacting with a giant mural while creating a fresh, original, visually striking composition.
Cinematic Rock Performance
Made with Seedance 2.5
Duration: 30 seconds
Aspect ratio: 16:9
Style: Photorealistic cinematic live action
Frame rate: 25 fps with natural 1/50-second shutter motion blur
Music: Original alternative rock with electronic textures, approximately 100 BPM
CORE DIRECTION
Create an emotionally intense rock-performance film inspired by Linkin Park’s blend of alternative rock, electronic textures and forceful band performance. Feature an original six-member band with consistent identities.
Combine the uploaded reference video’s aging manor, intimate wide-angle portraits and muted vintage colors with powerful live-performance energy. Use original music and voices throughout; do not reproduce existing lyrics, melodies, samples or a specific singer’s vocal identity.
SETTING & CONTINUITY
An abandoned manor ballroom featuring cracked botanical wallpaper, dark wooden panels, worn parquet flooring, exposed plaster, tall windows and an adjoining narrow hallway.
A real drum kit stands at the rear. Guitar and bass amplifiers sit along the sides, with an electronic sampler on a stand. Instrument cables lie naturally across the floor.
Maintain consistent room geography, equipment placement, window positions and hallway access throughout every shot.
BAND — EXACTLY SIX MEMBERS
Lead vocalist: Short dark hair, lean build, charcoal T-shirt, dark trousers and worn black boots. Uses a handheld wired microphone.
Second vocalist / electronics performer: Cropped black hair, dark overshirt, grey T-shirt and black trousers.
Guitarist: Distinct face, understated dark clothing and the same guitar throughout.
Bassist: Distinct face, understated dark clothing and the same bass throughout.
Drummer: Distinct face, understated dark clothing, performing on the rear drum kit.
Sampler / turntable performer: Distinct face and understated dark clothing, positioned at the electronic station.
Preserve each member’s face, hairstyle, outfit, role and instrument across all shots.
TIMED SEQUENCE
00:00–00:05 — RESTRAINED TENSION
Open in an extreme close-up of the lead vocalist’s hand gripping the microphone. Its cable hangs with natural weight.
Tilt up to his face. His head is lowered; he breathes quietly. As he looks up, a cold window reflection catches his eyes.
Cut wide to reveal the band waiting in the dim ballroom. A faint electronic pulse begins. The drummer lifts his sticks into position.
00:05–00:10 — THE FIRST IMPACT
On a strong drum accent, the full band enters.
Use a low 18mm shot near the guitarist. His picking hand strikes the strings while his fretting hand holds a believable chord. The strap visibly supports the guitar’s weight.
Hard cut to the drummer landing a snare hit, followed by a cymbal strike. Show clear stick contact and realistic cymbal flex and vibration.
Push toward the lead vocalist as he steps forward and brings the microphone to his mouth.
00:10–00:15 — CLOSE, RHYTHMIC DELIVERY
The second vocalist delivers this original rhythmic English line:
“I kept my voice behind these walls.”
Track beside him at chest height as he takes two steps toward the lead. His gestures remain compact and rhythmically precise.
Cut to a tight portrait of the lead listening, jaw tense, then to the bassist plucking in time.
Keep all visible playing physically credible and synchronized with the music.
00:15–00:20 — QUIET FRACTURE
Drop the arrangement to filtered electronics and a sustained guitar texture.
Match-cut to the lead alone in the adjoining hallway, with the ballroom visible behind him. Use a slow push-in with a 35mm lens.
He sings softly in an original male voice:
“Now every crack is letting go.”
His expression shifts from restraint to resolve. He turns toward the ballroom. Cut on the turn.
00:20–00:25 — CHORUS RELEASE
Cut to a full-band wide shot as heavy guitars and forceful live drums return.
The lead sings with powerful, rough-edged intensity:
“I won’t disappear inside these walls.”
Use an original vocal identity. He bends his knees and leans into the phrase with grounded balance. The second vocalist adds a brief harmony using the same lyric, without introducing additional words.
Make a fast, physically plausible lateral camera move past an amplifier, revealing the whole band through foreground parallax.
Cut to the drummer on a strong accent.
00:25–00:30 — FINAL HIT
Alternate between two readable compositions:
A medium close-up of the lead finishing the chorus phrase.
A low full-band view with the guitarist and bassist moving to the rhythm.
On the final coordinated hit, the drummer strikes a cymbal, the guitarist and bassist stop their picking hands, and the lead lowers his microphone.
Rapidly pull back through the doorway, leaving the band framed within the ballroom.
Hold the final second as the cymbal vibrates and guitar feedback decays into natural room reverb.
CAMERA & PERFORMANCE REALISM
18mm: Low-angle performance shots.
24mm: Handheld proximity and intimate wide-angle portraits.
35mm: Emotional portraits and the hallway push-in.
Physical camera momentum, short push-ins, restrained Dutch angles and beat-driven hard cuts.
Preserve facial identities during movement.
Show real skin texture, slight perspiration, weighted clothing and firmly planted feet.
Maintain believable hand positions, instrument handling, microphone cable movement and drummer contact.
Keep impacts, picking, plucking and vocal delivery synchronized with the soundtrack.
LIGHTING & COLOR
Cold daylight enters through tall windows. Warm practical lamps provide subtle illumination at the room’s edges.
Use muted teal, charcoal, tobacco brown and dirty cream. Preserve deep, detailed shadows, gentle halation, subtle film grain and restrained saturation.
No concert lasers or glossy studio lighting.
ORIGINAL AUDIO
Build an original alternative-rock/electronic arrangement at approximately 100 BPM:
Opening: Sparse electronic pulse and quiet room ambience.
First impact: Distorted guitars, heavy bass and forceful live drums.
Rhythmic section: Tight instrumental support beneath the second vocalist.
Hallway section: Filtered electronics and sustained guitar texture.
Chorus: Full-band intensity with a brief supporting harmony.
Ending: A coordinated final hit, vibrating cymbal, natural room reverb and fading guitar feedback.
Use only these lyrics:
“I kept my voice behind these walls.”
“Now every crack is letting go.”
“I won’t disappear inside these walls.”
No existing Linkin Park lyrics, melodies, samples or cloned voices.
AVOID
Changing identities; swapped instruments; extra band members; incorrect finger placement; disconnected straps; floating cables; drumsticks passing through drums; unsynchronized impacts; duplicated limbs; plastic skin; impossible camera moves; excessive slow motion; captions; logos; watermarks.
Seedance 2.5
Prompt:
Create a 30-second photorealistic cinematic action short on a bright summer day in a Japanese park. A young adult Japanese woman wearing a school-style uniform moves through a surreal size-changing adventure. Realistic physics, high-speed movement, seamless scale transitions, dynamic camera work, natural sunlight, detailed environments, cinematic 4K. No text, subtitles, logos, or watermarks.
[0–5s] — THE JUMP
She crouches confidently on a thick tree branch in a sunny park, focused and fearless. Without hesitation, she jumps forward. The camera dives rapidly alongside her as she falls toward a picnic table below.
[5–9s] — THE SIZE SHIFT
Just before landing on the picnic table, her body suddenly shrinks to the size of a soda can. The tabletop transforms into a gigantic landscape filled with oversized sandwiches, fruit, chips, and drinks. She immediately starts running between the giant food items.
[9–13s] — THE BEE
A bee suddenly races past her like a speeding vehicle. She jumps onto its back and grabs onto its body. During the movement, another seamless scale transition makes her roughly the same size as the bee. The bee rapidly takes flight.
[13–18s] — HIGH-SPEED FLIGHT
The bee weaves through the park at incredible speed. She holds tightly onto its back as they race between pedestrians' legs. Giant shoes hit the ground around them like moving obstacles. Strong wind blows her hair and clothing backward while the camera tracks them at ground level.
[18–22s] — WATER BALLOON CHAOS
They fly through a group of children having a water-balloon fight. Time suddenly shifts into dramatic slow motion. Water balloons burst around her, with thousands of sparkling droplets suspended in the sunlight. She looks both nervous and thrilled as the bee narrowly avoids the explosions.
[22–25s] — HUMAN TUNNEL
Speed instantly returns to normal. The bee darts through a line of cheerleaders standing side by side. Their legs and colorful pom-poms create a rhythmic tunnel around the tiny rider. The camera races alongside them with energetic tracking movement.
[25–27s] — THE TRANSFORMATION
She jumps from the bee. While falling, her body rapidly expands back to normal human size in one seamless visual transition. A young adult male student nearby opens his arms and catches her safely.
[27–30s] — FINAL MOMENT
Both are now normal-sized. She briefly hugs him, then turns toward the camera with a playful smile and a quick wink. He looks surprised and delighted. Freeze briefly on the charming final frame.
STYLE: photorealistic cinematic action, realistic human anatomy, natural facial expressions, physically believable movement, seamless scale transformations, bright summer sunlight, realistic shadows, detailed park environment, dynamic FPV-style camera movement, controlled motion blur during high-speed transitions, slow-motion water droplets, cinematic depth of field, 4K HDR.
Everyday Worldbuilding Vignette
You can think of this as a general-purpose prompt. Based on your character, it creates a series of everyday moments that explore their world, environment, routines and the people and systems around them.
You can try it with any visual style too, whether it's anime, realistic, 3D or something different.
I created this video with Seedance 2.5 but the prompt also works with Wan 3.0. It should work with H3 and other video models too. You may just need to adjust the number of cuts depending on the model and video duration. You can try it in any
What can you use it for?
- Explore and expand your character's world
- Grab frames from individual scenes to build environment references
- Cut out certain shots and use them as B-roll
- Use the footage in music videos
It's also a fun way to let the model interpret your character and see what kind of world it builds around them.
You can find the prompt below 👇
GPT Image 2.5 is crazy for storytelling..
give it a simple idea and it can create the storyboard, then animate it with Seedance 2.5
free workflow + prompts below 👇