Seedance 2.5 via OpenArt Beginner 15 min Free

Superhero Landing — You Fall From the Sky and Talk to Camera

Two prompts: a character sheet of you, then a 10-second video where you land in a Brooklyn street, say your own line, and the bridge explodes behind you.

Generated end to end. Sound is part of the render. Press play to watch.
The short way

Have it running in three steps

  1. 1
    Copy the prompt

    It is right above. No signup needed to take it.

  2. 2
    Open OpenArt

    Daily free credits. Signing up takes about a minute.

  3. 3
    Paste it and generate

    Takes about a few minutes. Then come back for the detail below.

Open OpenArt ↗

What you'll need

15 min
  • OpenArt

    Many image and video models behind one account, with a playground for comparing them.

    Daily free credits

A woman drops out of the sky. She lands in a crouch on a cobblestone street in Brooklyn. She stands up, looks into the camera and says one line. Then the bridge behind her blows up, and she does not even turn around.

It is a strong hook for a Reel. The person in it can be you, and the line can be anything you want to say.

You need two prompts. The first one makes a character sheet: one image of you from every side. The second one makes the video from that sheet.

The whole process

  1. Generate a character sheet with the first prompt. Use a photo of your face and a photo of an outfit.
  2. Open OpenArt, go to Video and pick Seedance 2.5.
  3. Add the character sheet as the reference image.
  4. Paste the video prompt. Change the line in { } to your own words.
  5. Press Generate.

The three stages

StageTimeWhat happens
10–3sShe falls from the sky and lands in a crouch. The stones crack.
23–7sShe stands up, looks into the lens and says the line.
37–10sThe bridge behind her explodes. She does not flinch.
01

Make the character sheet

You need two images. Image 1 is a clear photo of the face you want. Image 2 is a photo of the outfit. It can be a photo of a model in a shop. The prompt takes only the clothes from it, never the face.

Generate it in any image model that takes two references. Set the shape to 16:9.

A character reference sheet of a woman with long red ombre hair, a dark green tank top and wide black jeans: a large face portrait, front, side and back full-body views, two profile close-ups and a strip of detail shots.
Reference The character sheet — face, full body from three sides, details and colours
Character sheet prompt

Create an ultra-photorealistic professional CHARACTER REFERENCE SHEET of ONE adult person.

REFERENCE MAPPING: Image 1 = IDENTITY reference. Image 2 = OUTFIT reference.

IDENTITY — IMAGE 1: Use Image 1 as the only source for the character’s identity. Preserve the exact recognizable face, facial proportions, bone structure, eye shape and color, eyebrows, nose, lips, skin tone, apparent age, hairstyle, hair length, hair texture and hair color. Do not beautify, redesign, stylize or replace the face. The same exact character must appear consistently in every panel.

OUTFIT — IMAGE 2: Recreate the complete outfit from Image 2 exactly. Match the same garments, cut, fit, proportions, fabric, texture, colors, shoes, jewelry and visible accessories. Do not copy the model’s face, body, hair or identity from Image 2. Transfer only the clothing and accessories onto the character from Image 1.

BODY CONSISTENCY: Use one consistent realistic body type and body proportions throughout the entire sheet. Keep the same height, shoulder width, waist, hips, arms and legs in every full-body view. Natural anatomy. No body reshaping between panels.

LAYOUT: Create one wide horizontal character sheet, approximately 16:9, composed of clean rectangular studio panels separated by narrow white gutters.

LEFT PANEL: A very large frontal head-and-shoulders portrait taking most of the left side. Straight-on camera angle. Neutral relaxed expression. Face clearly visible with high facial detail.

CENTER SECTION: Three full-body views placed side by side:

1. FULL BODY FRONT Standing naturally facing directly toward camera. Entire body visible from head to shoes. Arms relaxed naturally.

2. FULL BODY SIDE Exact 90-degree side profile. Entire body visible. Natural neutral posture.

3. FULL BODY BACK Exact rear view. Entire body visible. Show hairstyle and outfit construction from behind.

RIGHT COLUMN: Two medium close-up portraits stacked vertically:

TOP: 3/4 facial view, turned approximately 30–45 degrees from camera.

BOTTOM: Exact side-profile portrait, approximately 90 degrees.

BOTTOM DETAIL STRIP: Add a horizontal strip of isolated macro reference panels showing:

1. close-up of one eye and eyebrow 2. close-up of lips 3. close-up of hair texture and color 4. close-up of one important jewelry or accessory detail from the outfit 5. close-up of another accessory if present 6. close-up of the main upper-body fabric or material 7. close-up of waist / trousers / skirt construction 8. close-up of fastening, pocket or garment detail 9. close-up of one shoe

At the far right of the detail strip, include 5–6 small circular COLOR SWATCHES sampled from the character’s hair, skin and outfit.

VISUAL STYLE: High-end fashion character reference photography. Seamless pure white studio background. Soft large diffused studio lighting. Neutral color balance. Minimal soft grounding shadows underneath the feet. Realistic skin texture, pores, fine facial hair and natural imperfections. Realistic fabric texture and garment construction. Sharp facial details. Consistent exposure and white balance across every panel. 85mm portrait-lens look for facial close-ups. 50mm natural perspective for full-body views. No wide-angle distortion.

CONSISTENCY IS CRITICAL: The exact same character must appear in every panel. Same face. Same apparent age. Same hairstyle. Same hair color. Same makeup or natural face styling. Same outfit. Same accessories. Same body proportions.

The full-body views must show the outfit accurately from front, side and back. Do not invent hidden garment details when they are visible in the outfit reference.

FINAL IMAGE: Finished polished character sheet only. Clean rectangular grid. No captions. No panel names. No written measurements. No logos. No watermark. No duplicate people. No extra limbs. No cropped feet. No cropped full-body figures. No props unless they are part of the outfit reference. No studio equipment visible. No environment other than the white studio background.

Upload the face photo first and the outfit photo second.

Check the sheet before you move on. The face must look like the photo, and the feet must be in every full-body view. If something is off here, it will be off in the video too.

02

Open Seedance 2.5 and add the sheet

Open OpenArt, go to the video page and pick Seedance 2.5. Upload the character sheet as the reference image. It is the only reference. The prompt calls it @Image1.

If OpenArt gives the image its own tag when you upload it, put that tag in place of @Image1 in the prompt.

03

Paste the prompt and change the line

The line she says is inside curly brackets in stage 2: {This is the landing effect, and this is how you do it.} Replace the words, keep the brackets. Keep it short. It has about three seconds.

Video prompt

[Generation Goal] Generate a 10-second photoreal live-action shot. The central subject is one young woman, the same as in the reference image. The primary event: she falls from the sky into a cobblestone Brooklyn street, lands in a superhero crouch directly in front of camera, rises, speaks one line to the lens, and the bridge behind her erupts in an explosion.

[Reference Materials] @Image1 defines the woman completely — face, skin, body proportions, hairstyle, hair colour and full outfit. Her appearance and clothing are exactly the same as in the reference image and never change. Do not use the white studio background, the flat studio lighting, the standing catalogue pose, or the frontal full-body framing from this image. This is the only reference material. Exactly one woman appears in the entire video — do not duplicate or add characters. Environment, camera, lighting and color grade come from this text only.

[Scene] A narrow cobblestone street in DUMBO, Brooklyn, walled on both sides by tall 19th-century red-brick warehouse buildings with rows of tall windows, black fire escapes and metal scaffolding. The street runs straight away from camera and is closed at its end by the Manhattan Bridge: a steel suspension bridge whose grey stone-and-steel tower and truss deck span the gap between the two brick buildings, suspension cables fanning down on both sides. Through the arched opening under the bridge, a distant Manhattan skyline with a slim art-deco skyscraper is visible. Low green trees and street furniture line both sidewalks. Damp cobblestones. Empty street — no pedestrians, no vehicles.

[Visual Style] Photoreal live-action, full-frame sensor, wide cinema prime, shallow depth of field. Color grade is teal and amber: deep desaturated petrol-teal in the shadows and sky, warm amber-orange in the highlights and fire, crushed blacks, dense film-like contrast, fine grain. The outfit colours from @Image1 stay true and unshifted under every light. Real skin texture with visible pores and flyaway hairs, natural motion blur, airborne dust and grit, faint handheld micro-shake, lens flare only from the fire. Not illustrated, not stylised, not CGI-looking surfaces: instead the texture of documentary footage shot on a real street.

[Stage 1 — 0-3 seconds — descent and impact] Initial state: empty cobblestone street, bridge intact at the far end, flat teal overcast daylight. Camera: low wide shot at cobblestone level, tilted up along the street toward the bridge; the camera tilts down slowly as the figure grows. Primary event: a small figure appears high in the sky above the bridge deck and drops toward camera, hair and clothing streaming upward in the air rush; at 3 seconds she strikes the cobblestones directly in front of camera in a superhero crouch — one fist and one knee down, head lowered. Physics: the impact drives a ring of dust and grit outward across the stones, cobblestones crack in a radial pattern under her fist, loose paper and leaves lift and settle. Audio: <a rising rush of wind>, <a heavy thud of impact>, <stone cracking>. End state: she is crouched in the foreground centre of the street, fist on the stones, dust hanging around her, bridge still intact behind her.

[Stage 2 — 3-7 seconds — rise and line] Continue from the previous stage: same woman, same appearance and clothing as in @Image1, same crouch on the cracked cobblestones, dust still settling. Camera: slow push-in from wide to medium, staying low-angle. Primary event: she rises from the crouch — chest first, then shoulders, head lifting last — brushes dust off her hand, brings her chin level and looks straight into the lens with a calm, controlled half-smirk, one eyebrow slightly lifted, then speaks. She rises between 3 and 4 seconds and speaks from roughly 4 to 7 seconds. Dialogue language: American English. She speaks in relaxed conversational American English, calm and confident, slightly amused: {This is the landing effect, and this is how you do it.} Physics: low dust still drifting around her feet, a few grains falling from her fingers as she brushes her hand, hair settling against her shoulders. Audio: <grit settling on stone>, <a hand brushing dust off skin>; her voice close and dry with a light reflection off the brick walls. End state: she stands at full height in the foreground centre of the street, hands loose at her sides, looking straight into the lens, cracked cobblestones at her feet, bridge still intact behind her.

[Stage 3 — 7-10 seconds — the bridge erupts] Continue from the previous stage: same woman, same appearance and clothing as in @Image1, standing at full height in the foreground centre of the street, facing camera, cracked cobblestones at her feet. Camera: holds the low medium framing with faint handheld micro-shake, then takes one short hard jolt at the moment of detonation. Primary event: behind her, at the far end of the street, the Manhattan Bridge deck erupts — a rolling amber fireball climbs through the steel truss, black smoke boils sideways into the gap between the two brick buildings, suspension cables whip loose. A shockwave rolls down the street toward camera: dust lifts off the cobblestones, her hair and clothing blow forward, loose paper tumbles past her. She does not turn, does not flinch and does not look back — she holds eye contact with the lens, half-smirk unchanged, blinking once. Lighting: the flat teal daylight is overtaken from behind by warm amber firelight, laying a hard orange rim along her shoulders, hair and legs while her front stays cool teal; one lens flare from the fire. Audio: <a deep explosion boom>, <steel groaning and snapping>, <debris raining onto stone>, a low ringing decay underneath. End state: she stands in the foreground centre of frame looking into the lens, rim-lit orange, the bridge burning behind her, smoke filling the gap between the buildings.

[Maintain Consistency] One woman only, in every stage — no second figure, no bystanders, no vehicles. Her face, hair and full outfit stay exactly as in @Image1 from the first frame to the last. She stays in the foreground centre of the street facing camera throughout; the bridge stays at the far end behind her, never beside or in front of her. The outfit colours stay true under every lighting condition, including the amber firelight. The camera never crosses behind her and never reveals the street behind camera. No subtitles, captions or on-screen text at any point.

One block, one copy. Change only the words inside the curly brackets.

Set the length to 10s and the ratio to 9:16 in the settings, then press Generate.

Why it works

The sheet is used for the person only. The prompt tells the model to take the face and the clothes, and to ignore the white studio, the flat light and the standing pose. Without that line, the model tries to bring the studio into the street.

Every stage ends with a clear picture. Each stage says where she is when it is over. The next stage starts from there. That is why the landing, the line and the explosion feel like one shot and not three clips.

She does not react. The explosion is loud, but she keeps looking into the lens. The prompt says it three ways: does not turn, does not flinch, does not look back. That calm is the joke, and it is what makes people watch again.

Where it goes wrong

What you gotThe fix
The face does not look like the photoRegenerate the character sheet first. The video copies the sheet
A white studio in the videoKeep the line “Do not use the white studio background”
Two women appearKeep “Exactly one woman” and “One woman only, in every stage”
She turns to look at the fireKeep “She does not turn, does not flinch and does not look back”
The line is cut offMake the line shorter. About ten words fit
Subtitles on screenKeep the last sentence about no on-screen text
A short or landscape videoSet 10s and 9:16 in the settings, not only in the text

Ready? Open OpenArt and start with the character sheet.

Why OpenArt

Why this workflow runs here

Many models, one account

The image and video models this workflow needs sit behind a single login.

Free credits every day

You get daily free credits, so you can try the prompt before you pay anything.

Compare before you commit

Run the same prompt on a few models in the playground and keep the best one.

Where it is weaker: credits go fast once you start generating video.

Make yours in the next 15 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
Open OpenArt — free to start ↗

Referral link — if you subscribe I earn a commission, at no extra cost to you. More on that.

Related prompts