Escalator Outfit Change AI Prompt: 3 Looks, 1 Shot
Escalator outfit change AI prompt for Seedance 2.5. You ride up an endless escalator three times, in three outfits, in one 13-second shot.
Have it running in three steps
- 1
- 2Open OpenArt
- 3Paste it and generate
What you'll need
30 min- OpenArt
The camera hangs above an escalator and looks down. The steps keep moving up. She rides past sitting on the steps, in a striped look. As she leaves at the top, she comes back in at the bottom in a floral kimono. Then once more, in a gold python look.
This escalator outfit change AI prompt makes it in one 13-second shot. There are no cuts and the camera never moves, because the escalator does the editing.
You need four pictures: one of your face and three outfits. This page gives you the video prompt we used, a prompt to make neat outfit pictures, and the step where you fit the prompt to your own looks.
The whole process
- Make a character sheet of yourself. It only sets your face and hair.
- Pick three outfits and turn each into a clean flat lay on white.
- Give the video prompt and your four pictures to Claude. Ask it to rewrite the face and outfit lines.
- Open OpenArt, go to video and pick Seedance 2.5.
- Add the character sheet first, then the three outfits in order.
- Set 13 seconds and 9:16. Paste your prompt and generate.
What happens in 13 seconds
| Time | What happens | Look |
|---|---|---|
| 0–1s | Empty steps moving up. Then a red beret comes in at the bottom | Look 1 |
| 1–5s | She rides up sitting on a diagonal, one leg along the steps. She lowers her sunglasses and looks over them | Look 1 |
| 5–6s | Look 1 leaves at the top. She comes in at the bottom in the kimono | Look 1 → 2 |
| 6–8s | She leans back on one arm, one knee up. The sunglasses sit on her headscarf | Look 2 |
| 8–9s | Look 2 leaves at the top. She comes in at the bottom in the knit sweater | Look 2 → 3 |
| 9–13s | She lies back on the steps, gold earrings on, and keeps rising as the video ends | Look 3 |
The four images, in order
The prompt points at these slots by number. Upload them in this order, or the model dresses her in the wrong look.
| Slot | What it is | What it controls |
|---|---|---|
@Image1 | Character sheet | Her face, hair and body. Never her clothes |
@Image2 | Outfit flat lay | Look one |
@Image3 | Outfit flat lay | Look two |
@Image4 | Outfit flat lay | Look three |
Make your character sheet
The camera is high and she is sitting, so her face is small in the frame. A character sheet gives the model more to hold on to than one selfie: a big portrait, the body from three sides, and close-ups of eyes, lips and hair.
Here the sheet does one job only. It sets who she is. The green top, the black jeans and the white trainers on our sheet never appear in the video, and the prompt says so by name. Do the same with yours: list what your sheet is wearing, so the model knows to leave it out.
Turn three outfits into clean flat lays
Save three complete looks, from Pinterest or from your own wardrobe. Each one needs shoes. She sits with her legs out, so her feet are on screen the whole ride.
Pick three looks that do not share anything. If two looks have the same bag or the same shoes, they can bleed into each other.
Pictures from different places come in different sizes, angles and backgrounds. Run each look through GPT Image 2 with this prompt. Attach only the pieces of one look per run. You get three pictures in the same style and at the same scale.
Create a clean, premium fashion flat-lay of ONE complete outfit, using ONLY the clothing items shown in the attached reference images.
FORMAT & BACKGROUND: Vertical 2:3 portrait image. Pure seamless white background (#FFFFFF), evenly lit, no texture, no props, no floor, no surface edges, no shadows except a very soft, subtle contact shadow under each item. Shot from directly overhead, perfectly flat top-down view, no perspective distortion.
LAYOUT (keep this exact layout for every outfit): Arrange the items as if they were worn by an invisible person, top to bottom, centered on the vertical axis: - top zone: headwear / sunglasses / jewelry (if present) - upper-middle: outerwear and top, laid flat, sleeves neatly arranged - middle: belt / bag placed beside the waist area - lower-middle: trousers / skirt / shorts, laid flat, full length visible - bottom zone: the pair of shoes, side by side, fully visible Even spacing between items, nothing overlapping more than slightly, nothing cut off by the frame. Generous white margins on all sides. The whole outfit fills about 80% of the frame height.
ITEM ACCURACY: Reproduce every item exactly as in the references: same colors, fabric, texture, cut, length, silhouette, prints, logos, hardware, stitching and details. Do not redesign, recolor, simplify or “improve” anything. Do not add any clothing or accessories that are not in the references. Exception: if no footwear is shown, add one simple pair of shoes that matches the style and color palette of the outfit. If an item in the reference is worn by a person, extract only the garment. No people, no body parts, no mannequins, no hangers.
STYLE: High-end e-commerce / fashion editorial flat-lay. Crisp focus edge to edge, true-to-life colors, soft diffused studio light, fabrics neatly pressed and smoothed, realistic fabric folds and material texture. No text, no labels, no watermark, no logo overlays, no frames, no collage borders.
Have Claude rewrite the face and outfit lines
The prompt below describes our woman and our three outfits in words. If you upload your pictures and leave our words in, the model gets two answers and mixes them.
Open Claude. Attach your character sheet and your three flat lays. Paste the whole prompt from step four. Then send this:
Here is a video prompt, my character sheet and my three outfit flat lays.
Rewrite only these parts so they describe my pictures instead of the ones written there now: 1. In block 01, the lines that describe her face and hair, and the list under CHARACTER SHEET WARDROBE IS FORBIDDEN (name what I wear on my sheet). 2. In block 02, the LOOK ONE, LOOK TWO and LOOK THREE item lists. Name every piece with colour and fabric, shoes included. 3. In block 06, the one gesture for each pose, so it uses something from that look. If a look has sunglasses, make sure the face is still visible. 4. In blocks 07 and 09, any word that names our hair, freckles or fabrics.
Change nothing else. Keep the escalator, the camera, the timings, the spacing between rides, the sound rules and every other line exactly as they are.
Give me back the full prompt in one block, ready to copy.
Check two things in what Claude gives you. Every look must name its shoes. And the forbidden list must name the clothes on your sheet, not ours.
Set it up and generate
Open OpenArt, go to video and pick Seedance 2.5.
Add the character sheet first, then the three flat lays in order. Set the length to 13 seconds and the shape to 9:16.
Paste your rewritten prompt. Do not shorten it. The timings in block 08 are made for 13 seconds, so a different length breaks the spacing.
INFINITE ESCALATOR — PREMIUM CINEMATIC FASHION EDITORIAL ONE PERMANENT CHARACTER IDENTITY THREE STRICTLY ISOLATED OUTFITS THREE DISTINCT FASHION POSES NATURAL DIEGETIC SOUND ONLY
Generate a premium, photorealistic, 13-second luxury fashion editorial film.
OUTPUT FORMAT: Vertical 9:16. Target 4K Ultra HD, 2160 × 3840. Maximum available native rendering quality. 30 FPS if supported. One uninterrupted continuous camera shot. One realistically moving escalator. One recurring human character. Three successive appearances in three different outfits. Natural real-world environmental sounds only. NO MUSIC OR AUDIO TRACKS OF ANY KIND.
CORE CONCEPT: A single recognizable woman appears three times on a continuously moving escalator, wearing a different accurately referenced outfit during each appearance. The identity of the woman never changes. The first, second and third appearances feature the exact same face, hair, skin, recognizable anatomical characteristics and natural body proportions. Only the clothing, accessories, fashion pose and position along the moving escalator change. The camera, environment, escalator speed and lighting remain consistent. The video feels like a genuinely filmed premium luxury fashion campaign.
━━━━━━━━━━━━━━━━━━━━━━━━ 01 — ABSOLUTE MASTER IDENTITY ━━━━━━━━━━━━━━━━━━━━━━━━ @Image1 = THE ONLY HUMAN IDENTITY REFERENCE. @Image1 is a multi-panel character sheet of ONE woman: a large front face close-up, full-body front, side and back views, two profile portraits and detail crops of her eyes, lips, hair and skin. Use all panels of @Image1 together as the sole source for the recurring character’s appearance. Preserve her recognizable face: light green eyes, soft natural brows, light freckles across the nose and cheeks, full natural lips, defined cheekbones, slim jawline and chin, fair skin tone. Preserve her hair exactly: very long, voluminous loose waves, center part, ombré from dark burgundy-black roots through deep red to bright copper-orange ends. Preserve her apparent age, slim build and natural body proportions. The exact same woman from @Image1 appears in all three outfits. This is ONE recurring individual, not three different people with a similar appearance. Do not independently reinterpret the face for each outfit. Do not generate a generic fashion-model face. Do not combine multiple identities. Do not borrow any human identity features from wardrobe references.
CHARACTER SHEET WARDROBE IS FORBIDDEN: The clothing and accessories worn in @Image1 are NOT wardrobe references and must NEVER appear in the video. Never show: the dark green ribbed sleeveless crop tank top, the black wide-leg denim jeans, the white low-top sneakers, the gold rectangular wristwatch, the small gold hoop earrings. The white background and the color-swatch dots in @Image1 are not environment or styling references. @Image1 provides ONLY face, hair, skin and body. Everything she wears comes exclusively from @Image2, @Image3 and @Image4.
━━━━━━━━━━━━━━━━━━━━━━━━ 02 — STRICT WARDROBE REFERENCE SYSTEM ━━━━━━━━━━━━━━━━━━━━━━━━ @Image2 = LOOK ONE ONLY. @Image3 = LOOK TWO ONLY. @Image4 = LOOK THREE ONLY. Each outfit reference is a flat-lay on a white background and an independent, closed wardrobe package. Assemble the flat-lay items into one complete wearable outfit, worn naturally by the woman from @Image1.
LOOK ONE — @Image2: Red wool beret with a silver round brooch, dark wraparound sunglasses, red-and-black leather choker with a silver O-ring, fitted red mesh t-shirt with a large printed hare graphic and blue ringer trims, wide-leg trousers with bold vertical beige-and-black stripes, black slouchy leather shoulder bag with chains and a small brown plush bear charm, chunky red-yellow-black velcro sneakers.
LOOK TWO — @Image3: Turquoise floral headscarf tied as a headband with a knot on top, blue-lens tortoiseshell cat-eye sunglasses, long open turquoise kimono with large orange-and-cream hibiscus print, mustard-gold satin twist bandeau top, ochre paperbag-waist trousers with white drawstring and rolled cuffs, small blue crossbody bag with colorful patchwork flap and gold clasp, blue suede criss-cross flat slides.
LOOK THREE — @Image4: Large hammered gold disc drop earrings, chunky gold disc collar necklace, fluffy cream mohair crewneck sweater with red and black scalloped fan pattern, high-waisted pleated wide-leg trousers in metallic gold python print, gold metallic square-toe flat mules.
OUTFIT EXTRACTION: Preserve the exact visible colors, prints, silhouettes, garment lengths, materials, fabric textures, cuts, layering and accessory combinations. Never invent replacement garments — each reference already contains a complete outfit including footwear.
STRICT OUTFIT ISOLATION: Look One must not contain garments or accessories from @Image3 or @Image4. Look Two must not contain garments or accessories from @Image2 or @Image4. Look Three must not contain garments or accessories from @Image2 or @Image3. No look contains anything from the character sheet @Image1. Never swap outfits. Never mix wardrobe references. The order is permanent: @Image2 → @Image3 → @Image4.
━━━━━━━━━━━━━━━━━━━━━━━━ 03 — PERMANENT ESCALATOR ENVIRONMENT ━━━━━━━━━━━━━━━━━━━━━━━━ Create a realistic modern indoor inclined escalator. The escalator is a physically ascending mechanical staircase, not a horizontal moving walkway. The central moving staircase contains precisely constructed silver-gray aluminum steps with dense parallel ribbed grooves, dark recessed risers, narrow black mechanical gaps and thin yellow safety markings along both sides. The metallic steps have realistic industrial texture, subtle wear, dimensional shadows and controlled reflections. Both sides contain brushed stainless-steel panels, dark handrails and transparent glass balustrades. The escalator runs vertically through the 9:16 composition. The left and right handrails establish the visual boundaries of the frame. Do not reveal large surrounding areas beyond the handrails. Keep the complete moving staircase, both side handrails and narrow adjacent structural elements visible. The escalator extends beyond the upper and lower image boundaries, creating the impression of an endless fashion conveyor. The environment remains identical throughout the video. No additional people, advertisements, signs or distracting background objects.
━━━━━━━━━━━━━━━━━━━━━━━━ 04 — FIXED CINEMATIC CAMERA ━━━━━━━━━━━━━━━━━━━━━━━━ One professional cinema camera is mounted above and slightly in front of the inclined escalator. Use a moderately elevated overhead-front angle. The perspective is slightly frontal rather than strictly vertical top-down. This angle allows the model’s face to be naturally visible from the beginning of each appearance without dramatic head movement. The camera is centered along the escalator axis. Position the camera slightly farther back than a tightly cropped overhead shot, but close enough to maintain recognizable facial details and clear wardrobe readability. Keep the full seated figure and the complete outfit, including footwear, visible during the main central portion of each appearance. The escalator handrails remain the left and right framing boundaries. Use realistic natural perspective without exaggerated wide-angle distortion. The camera remains rigidly fixed for the entire 13 seconds. No zoom. No pan. No tilt. No orbit. No tracking. No handheld shake. No reframing. No camera cuts.
━━━━━━━━━━━━━━━━━━━━━━━━ 05 — PHYSICALLY ACCURATE ESCALATOR MOVEMENT ━━━━━━━━━━━━━━━━━━━━━━━━ The escalator continuously moves UPWARD from the bottom of the image toward the top. Maintain a constant moderate speed throughout all 13 seconds. The movement is smooth, mechanically realistic and suitable for fashion presentation. Each metallic step retains its original geometry and spacing as it travels upward. The fixed side structures remain stationary. Each appearance of the model is physically seated on a specific moving step. The pelvis is naturally supported by the step. Legs, footwear, hands, bags and accessories interact realistically with the supporting steps. The model and the supporting steps move upward at exactly the same speed. The body remains attached to its supporting step as it passes through the frame. No independent vertical drifting. No floating. No hovering. No sliding across steps. No sudden changes of movement speed. Maintain realistic gravity, body weight, contact shadows and fabric compression.
ESCALATOR STEP CONTINUITY: Preserve the exact same physical escalator steps and their ribbed metallic textures continuously from the first frame to the last. Every step moves upward along one uninterrupted mechanical trajectory. Never regenerate, redraw, replace, reshape or reposition the escalator steps between consecutive frames. No texture popping. No step-pattern resets. No visual jumps in the escalator structure.
━━━━━━━━━━━━━━━━━━━━━━━━ 06 — THREE DISTINCT LUXURY FASHION POSES ━━━━━━━━━━━━━━━━━━━━━━━━ Each appearance has a clearly different premium fashion editorial pose. Every appearance enters already positioned in its assigned pose.
POSE ONE — DIAGONAL EDITORIAL (Look One): A relaxed diagonal seated pose. The torso is slightly rotated. One leg in striped trousers extends naturally along the steps, the other is bent, the chunky sneakers clearly visible. The black bag rests beside her on the step. Her one gesture: with two fingers she slowly lowers the wraparound sunglasses down her nose and looks over them into the camera, revealing her green eyes.
POSE TWO — ASYMMETRIC FASHION LEAN (Look Two): The torso leans gently backward with support from one arm. One knee raised, the other leg relaxed, creating a sculptural asymmetric silhouette. The long kimono drapes open across the steps. The cat-eye sunglasses are pushed up onto the headscarf so the face is fully visible. Her one gesture: the free hand lightly lifts the edge of the kimono and lets it fall.
POSE THREE — ELEGANT SEMI-RECLINED (Look Three): A semi-reclined editorial pose. The pelvis remains fully supported. Both elbows rest on the step behind her, legs in gold python trousers extended and loosely crossed at the ankles, gold mules visible. Her one gesture: a slow head tilt that makes the gold disc earrings swing and catch the light.
All poses remain anatomically correct, stable on the moving steps and compatible with the complete outfit. Allow small hand adjustments, natural breathing, realistic hair movement and realistic fabric motion. Avoid large pose changes during the upward journey.
━━━━━━━━━━━━━━━━━━━━━━━━ 07 — CONSISTENT FACIAL IDENTITY AND FASHION EXPRESSION ━━━━━━━━━━━━━━━━━━━━━━━━ Preserve the exact recognizable identity from @Image1 throughout the entire film. The face, green eyes, freckles, lips, jawline and the burgundy-to-copper ombré hair remain identical in all three appearances, including under the beret and the headscarf. The model’s identity does not change when the clothing changes. Preserve stable facial proportions between consecutive video frames. The face is naturally oriented toward the slightly angled camera from the beginning of each appearance. No dramatic head lifting. No abrupt head rotations. Allow subtle natural blinking, restrained gaze changes and quiet luxury-fashion expressions. No beauty blur. No artificial skin smoothing — keep her freckles. No invented facial features. No identity substitution. No temporal facial deformation. Use soft frontal fill lighting to keep the face readable.
━━━━━━━━━━━━━━━━━━━━━━━━ 08 — PRECISE 13-SECOND APPEARANCE STRUCTURE ━━━━━━━━━━━━━━━━━━━━━━━━ FIRST: @Image2. SECOND: @Image3. THIRD: @Image4. The next appearance begins entering only when the previous appearance has reached the upper region and begins disappearing beyond the top of the frame. Do not create a crowded queue of characters traveling closely together.
00:00–00:04.33 — LOOK ONE: The woman from @Image1 is already entering from the bottom of the frame, wearing only the complete outfit from @Image2, already in Pose One. She lowers her sunglasses and looks over them into the camera. By approximately 00:04.33 she approaches the upper edge.
00:04.33–00:06.50 — FIRST EXIT / SECOND ENTRANCE: The first appearance begins disappearing beyond the top edge. Only as this exit begins, the same woman enters from the bottom wearing the complete outfit from @Image3, already in Pose Two. The two appearances remain physically separated by several moving steps.
00:06.50–00:08.67 — LOOK TWO: The second appearance becomes the main subject. Only the outfit from @Image3. She lifts and releases the edge of the kimono. The face remains the same as in the first appearance.
00:08.67–00:10.83 — SECOND EXIT / THIRD ENTRANCE: The second appearance begins disappearing beyond the upper edge. Only then does the third appearance enter from the bottom, wearing only the outfit from @Image4, already in Pose Three.
00:10.83–00:13.00 — LOOK THREE: The third appearance becomes the main subject. She tilts her head and the gold disc earrings swing. The escalator continues carrying her upward as the shot ends. The composition suggests an endless, continuously operating fashion presentation.
━━━━━━━━━━━━━━━━━━━━━━━━ 09 — PREMIUM CINEMATIC VISUAL FINISH ━━━━━━━━━━━━━━━━━━━━━━━━ High-end luxury fashion advertising. Sophisticated editorial cinematography. Soft directional overhead illumination with gentle frontal fill. Moderate contrast and refined tonal separation. Consistent facial illumination. Natural skin texture with visible freckles. Accurate clothing colors and clearly visible garment construction: mesh print, bold stripes, leather bag, floral kimono, satin bandeau, fluffy mohair knit, metallic python texture, hammered gold jewelry reflections. Rich glossy hair with the red-to-copper ombré catching the light. Sharp silver metallic escalator grooves and realistic handrail reflections. Professional cinematic color grading. Subtle analog 35mm film grain. No excessive diffusion. No aggressive beauty filtering. No artificial oversharpening. Maintain stable visual quality throughout all 13 seconds.
━━━━━━━━━━━━━━━━━━━━━━━━ 10 — STRICT NATURAL SOUND POLICY ━━━━━━━━━━━━━━━━━━━━━━━━ ABSOLUTELY NO MUSIC OR PRE-RECORDED AUDIO TRACKS. No soundtrack, no background music, no melody, no beat, no percussion, no cinematic score, no singing, no vocals, no dialogue, no voiceover, no narration. The audio consists ONLY of natural sounds produced within the scene: the continuous soft mechanical hum of the escalator motor, subtle vibration and gentle metallic sounds of the moving steps, soft fabric rustling, the faint clink of the bag chains and gold jewelry, a very faint indoor room tone. Quiet, restrained, realistic, synchronized with the visible actions, like clean production audio recorded on location.
━━━━━━━━━━━━━━━━━━━━━━━━ 11 — FINAL GENERATION PRIORITIES ━━━━━━━━━━━━━━━━━━━━━━━━ PRIORITY ONE: One single recurring woman, identity exclusively from @Image1. PRIORITY TWO: Exactly the same face and ombré hair across all three appearances. PRIORITY THREE: Strict wardrobe isolation — Look One = @Image2 only, Look Two = @Image3 only, Look Three = @Image4 only. Nothing she wears in @Image1 ever appears. PRIORITY FOUR: Three clearly distinct editorial poses, each with its one gesture. PRIORITY FIVE: Correct appearance spacing — the next appearance enters only when the previous one begins exiting above. PRIORITY SIX: Physically realistic seated body support, synchronized with the escalator steps. PRIORITY SEVEN: One stationary elevated overhead-front camera, handrails as lateral boundaries. PRIORITY EIGHT: Premium cinematic lighting, natural skin texture, subtle 35mm grain. PRIORITY NINE: Only natural scene sounds. NO MUSIC. NO AUDIO TRACKS. NO VOICES.
Why this prompt works
The camera is locked and only the escalator moves. The steps carry her out of the top of the frame, and that exit works as the cut.
The next ride starts only as the last one leaves. The prompt gives exact seconds for each exit and entrance. That keeps one of her on screen most of the time, two only for a moment, never three.
She sits on a step and moves with it. The prompt ties her body to one step and its speed. Without that, she floats or slides down while the steps go up.
Each outfit is a closed package. Look one may use nothing from looks two and three, and the reverse. The clothes on the character sheet are banned by name. That is what stops the green top from showing up on the escalator.
Each look gets one gesture: she lowers the sunglasses, lifts the kimono or tilts her head. One small move keeps her alive. With more than one, her hands and legs start to warp.
Where it goes wrong
| What you got | The fix |
|---|---|
| She wears the clothes from the character sheet | Name those clothes in the CHARACTER SHEET WARDROBE IS FORBIDDEN list |
| Pieces from one look show up in another | Keep the STRICT OUTFIT ISOLATION block, and pick looks that share nothing |
| Three of her ride up together like a queue | Keep block 08 and the line “the next appearance begins entering only when the previous appearance” |
| She floats or slides on the steps | Keep the lines “the model and the supporting steps move upward at exactly the same speed” |
| Her face looks different in one look | A hat or headscarf hides her hair. Keep the line “including under the beret and the headscarf” and use a sheet with a big clear face |
| Sunglasses hide her face the whole ride | Give that look a gesture that moves the glasses, like lowering them or pushing them up |
| The camera drifts or zooms | Keep block 04 with every “No zoom. No pan.” line |
| Music in the audio | Keep block 10. Add your own track later in the edit |
Questions people ask
Which model is this for?
Seedance 2.5. We ran it on OpenArt at 13 seconds and 9:16.
Can I put myself in it?
Yes. Make a character sheet from your own photos and use it as @Image1.
Then have Claude rewrite the face lines for you.
Does it have to be vertical?
The prompt is written for 9:16. The escalator runs from top to bottom of the frame, so a wide shape leaves empty space on the sides.
Is the prompt free?
Yes, everything on this page is free to copy. Generating the video uses credits on the platform.
Can I use more than three outfits?
Not in 13 seconds. Each look needs about four seconds to ride through. For more looks, make the video longer and add a block for each new look in block 08.
Ready? Open OpenArt, add your four pictures and paste your prompt.
Why this workflow runs here
Many models, one account
Free credits every day
Compare before you commit
Make yours in the next 30 min
You have the whole prompt and every setting. The only thing left is an account, and that takes a minute.
- Free to start
- Prompts ready to paste
- Upgrade only if you want more
Referral link — if you subscribe I earn a commission, at no extra cost to you. More on that.
Related prompts
Effect Free 24 GTR AI Prompt: Swap the Car for Your Dragon
A 24 GTR AI prompt for the AIReel effect. You flash the UNO cards, the scene cuts to a stone mountain, and a dragon stands behind you instead of a car.
Effect Free Fan Blade Outfit Change: The Top-Down AI Prompt
A free Seedance 2.5 prompt. She lies on a rug, a slow ceiling fan blade sweeps over her, and her outfit is new. Five looks, 20 seconds, five reference images.
Effect Free Fit Check Hit by a Car: The AI Video Prompt
A free Seedance 2.5 prompt. She films a street fit check on her phone, shows the outfit from every side, and a car takes her out of frame. 12 seconds, one reference image.