Paper RainAI filmmaking • guides • little discoveries

KEY LEE CREATES · SUPERHERO VIDEO GUIDE

How to Recreate Spiderman

Build the character. Let gravity go. Follow the swing.

The superhero reveal works because the character is clear before the action begins.

This workflow uses a selfie to establish identity, a character sheet to lock the costume, and one long motion prompt to control the fall, web swing, wall run and final city flight.

The concept is adapted from the original How to make the viral Gwen Stacy video post. The prompts below are an original Paper Rain interpretation for Key Lee Creates.

The result is a Spider-Man-inspired transformation experiment, not an official studio production. Use your own image and adapt the costume, setting and line to your story.

Paper Rain prompt card for the How to Recreate Spiderman guide
Create This in Higgsfield ↗

Affiliate disclosure: If you sign up through this link, I may earn a commission at no extra cost to you. I only share tools I actually use or find useful.

The workflow

Build the still character first. Then use that character sheet with the selfie as the identity reference for the video.

SELFIECHARACTER SHEETMOTIONEDIT

The important order is simple:

  1. Generate a three-view character sheet from your selfie.
  2. Check the face, costume layers and shoes before moving on.
  3. Use the selfie and character sheet together for the video.
  4. Give the video model one continuous action with explicit body physics.

Stage 1 — build the character sheet

Open the image-generation area in Higgsfield. Add your selfie as the image reference and use the prompt below. Replace the bracketed identity description with the person in your reference.

Keep the three views in one clean image: a full-body left view, a dynamic center view and a portrait right view. The sheet gives the video model a stable identity, outfit and body reference.

IMAGE PROMPT

CHARACTER SHEET — three views side by side, clean white studio background, professional superhero costume concept art for film production, photorealistic cinematic quality, sharp detail, dramatic studio lighting with subtle rim light and soft fill.

CHARACTER APPEARANCE:

[tag your selfie here] Describe the person accurately: age range, height, build, hair shape and color, eye shape, skin tone and any distinctive features. Preserve the person's identity and natural facial proportions.

The character is a slender, lean, athletic young adult with controlled gymnast and dancer-like movement. Keep the same face, body scale and hairstyle in every view.

SHE IS WEARING ALL OF THE FOLLOWING AT THE SAME TIME, ALL LAYERS VISIBLE IN EVERY VIEW:

1. BASE SUIT — a full-body, tight, seamless athletic suit covering the neck, arms, fingers and legs to the toes. It is fully opaque with no mesh, net, sheer panels or exposed skin below the neck. The suit has a clean neckline, full-length legs and integrated gloves with individually covered fingers.

Use clean graphic color zones: solid white on the outer arms, hands, fingers and front of the legs; dusty rose pink on the inner arms, palms, inner fingers, outer thighs and outer calves; solid matte black on the chest, torso, back and sides. Keep each color zone smooth, flat and free of patterns.

2. TOP GARMENT — a fitted dark navy knit school-style sweater with a white collared shirt underneath and a slim navy tie. The sweater has a plain, clean fabric with no badge, emblem or logo. Sleeves end at mid-forearm so the covered lower arms, wrists, hands and fingers remain visible.

3. WAIST ACCESSORY — an oversized washed-out heather-gray hoodie loosely tied around the hips, with the sleeves knotted at the front and the kangaroo pocket visible.

4. SHORTS — very baggy, oversized dark charcoal distressed cargo denim shorts worn over the suit. The hem falls below the knee. Include frayed edges, natural faded areas, small rips, heavy stitching and four large cargo pockets. The shorts must remain clearly visible over the hips and thighs.

5. SHOES — teal turquoise professional ballet pointe shoes with satin material and long ribbons wrapped around the ankles.

6. ACCESSORIES — large black over-ear professional headphones resting naturally around the neck.

LEFT VIEW — full body from head to toe, face visible, confident pose, complete outfit clearly separated into layers.

CENTER VIEW — full body in a dynamic stance, arms slightly raised, hands open, all fingers individually covered and visible.

RIGHT VIEW — portrait close-up at a three-quarter angle, one covered hand raised near the face, cinematic gaze, hair and facial details consistent with the selfie, white collar, navy tie and headphones visible.

Photorealistic cinematic color grading, realistic skin and fabric texture, production-quality costume concept photography, consistent identity and proportions across all three views. No text, no captions, no logo and no watermark.

Check the still before making the video

Do not rush into motion. The character sheet needs to be useful as a reference, not merely attractive.

  • The face should match your selfie.
  • The full outfit should be visible in the center view.
  • The hoodie, shorts, pointe shoes and headphones should not disappear.
  • The hands and fingers should be anatomically clear.
  • There should be no logos, captions or accidental text.

If the costume is inconsistent, simplify the description and regenerate the sheet before adding motion.

Stage 2 — make the video

Open the video-generation area. Add both your selfie and your character sheet. Tag them in the prompt where marked:

  • [tag your selfie here] keeps the face and identity consistent.
  • [tag character sheet] keeps the outfit and body design consistent.

Paste the motion prompt below. It is written for a single continuous 20-second shot. The camera, body positions, sound and environment are all called out because vague action prompts tend to create floating bodies, random cuts and changing costumes.

VIDEO / MOTION PROMPT

Seedance 2.5 — One Continuous 20-Second Shot

CHARACTER REFERENCES — MANDATORY:

Face and identity: [tag your selfie here].

Outfit and costume: [tag character sheet].

Preserve the same person, face, hairstyle, body scale, outfit layers, teal satin pointe shoes and black headphones in every frame.

CRITICAL — CHARACTER ANATOMY:

The character is a young adult woman with a slender, lean, athletic build and controlled gymnast/dancer movement. Keep proportions consistent in every frame. Joints bend naturally. The spine curves correctly. The core stays engaged. Never use ragdoll motion, limp limbs, floating limbs, broken joints or inconsistent scale. Every movement follows real physics.

CRITICAL — FACE VISIBILITY:

The face is fully visible only during the opening window moment. After the fall begins, follow the character from behind for the rest of the shot. A quick accidental profile is acceptable, but never show a direct frontal close-up after the window. Read emotion through body language and sound.

CRITICAL — PHYSICAL CONTACT:

The character interacts physically with the city. Feet push from walls, the body grazes fire escapes, and feet can bounce from car roofs. Every contact has weight: knees absorb impact, muscles engage and the body reacts. The character is parkouring and web-swinging, not floating.

CRITICAL — BODY POSITIONS:

Back-first fall from the window: calm, arms slightly out, eyes closed, peaceful face, relaxed body, hoodie and hair responding to gravity.

Web swing: arms extended forward gripping the web, legs together trailing, core tight, back straight and chin lifted.

Wall push-off: both feet plant flat, knees bend deeply, then the legs extend powerfully as the arms drive forward.

Running on a wall: body leans approximately 45 degrees toward the wall, feet strike in a running gait and arms pump.

Free-flight peak: arms open naturally, body slightly arched, physically released.

Street-level run: full sprint, arms driving, head forward and pointe shoes striking the pavement with believable weight.

CRITICAL — EMOTION THROUGH BODY:

The fall communicates calm surrender and peace. The first swing shows focus and a tight grip. The car-roof contact creates a quick surprised laugh. The wall run feels explosive and determined. The low taxi pass feels aggressive and controlled. At the words “WOOHOO!” the chest opens, arms spread and the body expresses pure joy. At the final peak the character becomes open, still and surrendered.

CRITICAL — SOUND:

No music and no score. Use sound effects only: city ambience, distant traffic, the windowsill creaking, the soft thud of headphones shifting, wind rushing, web snap and twang, feet contacting a car roof, wet pavement and a taxi horn. Include the casual opening line, a brief laugh and “WOOHOO!” echoing between buildings. End with wind fading into silence at the peak.

CRITICAL — ENVIRONMENT:

Real lived-in residential New York City. Begin outside a high-floor apartment window with a messy teenage bedroom visible through the glass: cluttered desk, unmade bed, posters, stacked books, energy drink cans and clothes on the floor. Below are brownstones, fire escapes, bodegas, pedestrians, a delivery worker and someone walking a dog. The city should feel inhabited, not like an empty tourist postcard.

CRITICAL — LIGHTING:

Early evening with soft, slightly overcast light. Warm light touches the building façades. The sky is natural gray-blue with a gentle pink horizon. Reflections of overcast brick buildings appear in the window glass. Keep the light subtle and believable, like a regular city evening.

CRITICAL — REALISM:

Photorealistic Hollywood action film. RED Monstro 8K look, anamorphic lens, real depth of field, natural lens flare, motion blur, fabric physics and atmospheric city haze. No AI artifacts, no floating bodies and no broken anatomy.

CRITICAL — CAMERA:

One continuous shot with zero cuts. The camera begins static outside the apartment, level with the window and facing the character through the glass. It does not move during the bedroom moment.

When the character falls, hold on the empty window for half a beat, then dive straight down and become a physical drone-follow camera behind the character for the rest of the shot. Follow the body through space with tilts, banking and severe rolling during the swings. Stay wide enough to show both the character and the environment. The horizon may cant 40–70 degrees during the most aggressive movement.

TIMELINE:

0–3.5 seconds — Static exterior camera at the high-floor apartment window. The messy bedroom is visible inside. The character sits casually on the sill with feet dangling over the city and headphones on the shoulders. She glances back into the room, gives an easy smirk and says, “Alright. Let’s do this one more time.” She makes a lazy one-finger spin.

3.5–5.5 seconds — She pulls the headphones over her ears. Without jumping or somersaulting, she leans back and falls back-first out of the window. She is calm, eyes closed and physically relaxed. The hoodie flows upward and the hair responds to gravity. She drops out of frame. Hold on the empty window for half a beat with wind and silence.

5.5–7 seconds — The camera plunges straight down and catches her mid-fall from behind. Move close to one open, white-and-pink gloved hand. The wrist snaps. A web shoots forward, snaps taut and vibrates realistically. The fall converts into a swing. Pull back quickly and lock the camera behind her.

7–10.5 seconds — Follow her low above a residential street with brownstones, fire escapes, bodegas and pedestrians. She swings approximately three meters above the street with arms extended, legs trailing, core tight and back straight. She passes close to a fire escape, instinctively tucks, then her pointe shoes skim a parked car roof. A brief surprised laugh is heard. The camera banks hard and the horizon cants.

10.5–13 seconds — Release at the swing peak. The body opens naturally, then runs along a brownstone wall at a 45-degree lean for four or five quick steps. Both feet push off hard with bent knees and full extension. A new web fires across a wide intersection. The camera rolls sharply as the swing yanks sideways and drops low between yellow taxis.

13–15.5 seconds — Move fast and low between taxis, approximately two meters above the ground. The body is streamlined with arms forward and legs trailing. Release the web. Feet strike wet pavement and the character sprints for three or four steps. A taxi honks and the character laughs. Plant one foot against a taxi for an explosive push-off, fire a web toward a tall building and rocket upward from directly behind. “WOOHOO!” echoes between the buildings.

15.5–18.5 seconds — High above the city, follow from behind in a wide aerial view. Show the regular New York grid, rooftops, water towers and fire escapes. Fire a third web and swing in a long arc. Bank severely until the horizon passes 70 degrees. At the peak, arms open, head back and body arches into one breathless moment of freedom. Show only the silhouette and the city below.

18.5–20 seconds — Drift behind the character mid-swing. The city stretches forward and the character becomes small against it, still moving toward somewhere.

ABSOLUTE CONSTRAINTS:

One single continuous unbroken 20-second take. No cuts, hidden cuts, edits, transitions, fades, camera resets, time jumps or discontinuous camera motion.

No distant establishing shot, random geometry, random costume changes, duplicated characters, impossible proportions, floating bodies, broken joints or camera teleportation.

No text, captions, logos or watermark.

The complete sequence must feel like one physical camera movement: a calm window moment, a back-first fall, a camera dive, a web swing, wall contact, a street run and a final city flight.

[tag your selfie here] [tag character sheet]

Generate and edit

Start with one generation so you can see which instruction needs work. Do not change every variable at once.

  1. Generate the character sheet and save the version with the clearest identity.
  2. Upload the selfie and character sheet to the video tool.
  3. Choose the available Seedance 2.5 option and a 20-second duration if your account offers it.
  4. Generate the full one-take video.
  5. Edit only after the motion works: trim the opening, balance the sound effects and keep the strongest final swing.

If the generator does not offer the exact model or duration, use the closest available option and preserve the prompt’s sequence and constraints.

If the superhero breaks

The face changes after the fall

Keep the selfie and character sheet attached together, repeat that the camera follows from behind after the window, and remove extra face descriptions from the motion prompt.

The costume loses layers

Regenerate the character sheet first. Make the hoodie, shorts, pointe shoes and headphones visible in the center view before using it as a reference.

The character floats instead of swinging

Keep the physical-contact instructions. Name the wall push, car-roof skim, wet pavement impact and knee absorption as separate actions.

The model adds cuts

Repeat “one continuous shot” near the camera section and timeline. Remove any language that asks for montage, transition, flash or scene change.

The city looks empty

Keep the residential details and name several background actions: pedestrians, a delivery worker, a dog, taxis and a taxi horn.

The video adds music

Repeat “no music and no score” and list the sound effects you want instead.

Hi, I'm Key Lee.

I make things at the intersection of AI, filmmaking, technology, music and storytelling.

The strongest action prompts do not only describe the spectacle. They describe what the body weighs, what the camera can see and what the environment does in return.

Give the impossible moment a body. Then let it move. 🤍