Book Transformation AI Video Prompt: Dragon War
Book transformation AI video prompt for Seedance 2.5. You read in a park, and the world turns into a dragon war around you, then back. 30 seconds, one take.
Have it running in three steps
- 1
- 2Open Vosu
- 3Paste it and generate
What you'll need
30 min- Vosu
This book transformation AI video prompt starts with a quiet evening. You lie on a picnic blanket in a park, with iced coffee and a bag of oranges, and open a book. Then the world around you turns into the story.
The grass dies into cracked earth. The trees turn into black stone spires, and the blanket becomes a torn war banner. A dragon crashes down behind you, and two armies fight a few metres from your feet. You keep reading. When you close the book, everything flows back into the park.
It is 30 seconds in one take. You need one clear photo of yourself. This page has three prompts: one for a character sheet, one for a park picture to plan the look, and the video prompt.
The whole process
- Pick one clear photo of your face, in good light.
- Make a character sheet of you with the first prompt, in GPT Image 2.
- Optional: make a park picture with the second prompt, to see the scene before you spend video credits.
- Open Vosu, go to video and pick Seedance 2.5. Add the sheet as @Image1.
- Set 30 seconds and 9:16. Paste the video prompt and generate.
What happens in 30 seconds
| Time | What happens |
|---|---|
| 0–5s | Golden hour in the park. She lies on the blanket and opens the book |
| 5–10s | The grass dies from under her. The blanket turns into a banner, the trees into stone spires, the sky goes grey |
| 10–16s | A dragon flies over and lands behind her. The sky turns storm dark, a spear stands in the ground |
| 16–24s | Two armies fight around her, dozens of dragons above, fire on the field. The camera rises high |
| 24–30s | She closes the book. The grass, the blanket and the trees come back. Golden light again |
Make the character sheet
The video is 30 seconds long and the camera moves around her a lot. One photo is not enough to keep the face and hair the same from every side. So first make a character sheet: you from the front, the side and the back, with close ups of the face.
Open an image model that takes a reference picture. We used GPT Image 2 on Vosu. Add your photo, set the shape to 3:2 and paste this prompt. Your face and hair come from the photo. The clothes are written in the prompt, and they are the same clothes the video asks for.
Use Image 1 as the only source for the person’s identity. Create a photographic character reference sheet of this person.
IDENTITY — IMAGE 1: Keep exactly: face shape and bone structure, eye shape and eye color, eyebrows, nose, lips, skin tone, freckles and moles, age, body build, hair color, hair length and hair texture. Do not beautify, slim or redesign the person. The same person must be recognizable in every panel.
HAIR: Keep the hair color and length from Image 1. Worn loose, with a natural part, visible strands and flyaways.
CLOTHES:
- a plain black sleeveless top, fitted, soft cotton jersey
- faded light-blue straight-leg jeans, worn denim with natural creases
- bare feet No jewelry that is not visible in Image 1.
LAYOUT: One character sheet on a landscape 3:2 canvas, clean white studio background, thin white gutters between panels. Left: a large head-and-shoulders portrait, facing the camera. Center: three full-body views, front, side profile and back, head to feet fully visible. Right: two close-ups of the face, three-quarter view and side profile. Bottom row: small detail crops of the eye, the lips, the hair ends, the top and the jeans.
PHOTOGRAPHY: Real studio photograph, 85mm lens, soft even light, true-to-life color, real skin texture with pores, no airbrushing.
CONSTRAINTS: One person only. Same person, same clothes in every panel. No text, no labels, no numbers, no watermark, no illustrated or 3D look.
This is the sheet we used. Ours shows a green top, black jeans and sneakers, because it was made earlier for other videos. It still worked: the video prompt describes the clothes in words, and the model followed the words.
Check the hair before you move on. Long two-tone hair is the first thing to drift in a long video. If the colors change between panels, make the sheet again.
Optional: make the park picture
This step is not needed for the video. The video prompt describes the park in words, and the sheet is the only reference. But a picture of the park helps you see the scene and the light before you spend video credits on 30 seconds.
Use the same image model, add your character sheet, set 3:4 and paste this prompt.
Use Image 1 only for the woman’s face, hair and body. Do not use the white background, the layout or the clothes of the sheet.
A candid photo in a city park at golden hour. The woman from Image 1 lies on her back on a cream fringed picnic blanket on deep green grass, her head on a linen bolster pillow, holding an open hardcover book above her face and reading. She wears a black sleeveless top, faded light-blue jeans and is barefoot. Her long hair spreads over the pillow.
On the blanket next to her: a plastic cup of iced coffee with a straw and a net bag of oranges.
Tall leafy trees behind her, a low amber sun, long soft shadows on the grass, a few fallen leaves. Far in the background, soft out-of-focus people sitting on the grass.
Camera: 50mm lens at a low angle from her feet side, shallow depth of field, natural color, light film grain, real photograph.
No text, no watermark, no illustrated look.
This is the picture we made. The blanket, the pillow, the coffee and the oranges are the same objects the video prompt names.
Set it up and generate
Open Vosu, go to video and pick Seedance 2.5.
Add your character sheet as the only reference, so it is @Image1. Set the length to 30 seconds and the shape to 9:16. Paste the prompt below. It is the exact prompt that made the video above.
Elaria is the name of our character. You can leave it as it is, or change it to your own name everywhere in the prompt. The model only uses it as a label.
[Goal] 30s photorealistic live-action fantasy. Elaria lies on a picnic blanket in a park at golden hour, opens a book, and the world physically transforms around her into a dragon war — ground, objects, light and sky — while she stays in place reading. She closes the book and it changes back. She is inside the battlefield, not watching it from outside.
[Elaria] @Image1 — face, freckles, green-hazel eyes, long wavy hair, dark burgundy roots into copper-orange ends, centre part. Do not use the white studio background or the upright posture. Black sleeveless top, faded light-blue jeans, bare feet. She lies on her back the whole video — never sits up, never moves off her spot. Her face, hair, clothes and position are the ONLY things that never change.
[Park — start and end] Golden hour. Deep green grass, tall leafy trees, low amber sun, long shadows. On the grass: a cream fringed picnic blanket, a linen bolster pillow under her head, a plastic cup of iced coffee, a net bag of oranges. One thick dark hardcover book.
[Transformed world] Same space, changed object by object: grass dies into cracked grey earth, trees blacken into bare stone spires, the blanket becomes a torn mud-stained war banner, the pillow a leather bedroll, the cup a dented metal camp cup, the net bag a coarse twine one. Black storm above, orange ground fire, white lightning. Beyond her: two armies on scorched ground, burning towers, and near-black dragons with oxblood wings fighting through the storm.
[Rule] Every change starts at the book and at the ground under her body and spreads outward to the horizon — never the reverse. Nothing is fenced off; the ground she lies on changes too.
[Camera] Low, at ground level, inside the space. Foreground objects pass through: a spear, running legs, smoke. One camera move per stage.
0-5s — Warm park. She picks up the book beside her hip, opens it, starts reading. Camera glides low along the blanket past the coffee cup.
5-11s — She turns a page. Green drains from the grass starting under her shoulders and races outward. Blanket tears into a banner, pillow into a bedroll, cup tips and becomes metal, trees go black and bare, sky closes to grey. Camera pulls back low so dead earth spreads past her on both sides. She glances at the grey ground by her hand, then reads on. End: no green anywhere; she lies on a banner on cracked earth, same clothes, same pose.
11-18s — She turns a page. Two dragons collide overhead and one crashes down 15m away. The ground jolts under her; grit and ash hit her jeans and the page; her hair whips. It climbs away. Camera arcs low around her past a planted spear and smoke. Sky darkens to storm blue-black, underlit orange. She grips the book, brushes ash off the page, smiles slightly.
18-25s — She turns a page. Two armies collide and the fight runs through her space — soldiers sprint past a few metres from her head and feet, never looking at her. Dozens of dragons fight above. Darkest point: near-black storm, orange fire below. Camera rises once at the end into a high wide — she is a small figure lying in the middle of a burning field.
25-30s — She closes the book. Everything reverses from under her outward: green floods back from beneath her shoulders, the banner softens into the blanket, bedroll into pillow, cup upright, spires back into trees, storm drawn into the book, amber light returns. Camera descends to the same low view as 0-5s. She rests the book on her chest and breathes out. End: exactly the opening park — same trees, same objects, same positions. One ash flake lands on the cover. Her hair moves once in wind that is gone.
[Dragons] Two hind legs, wings as forelimbs, no front legs. Heavy shoulders, long tail, plausible wing structure. They move like heavy animals, not weightless CGI.
[Consistency] Elaria is the anchor — identical face, hair, clothes and pose in every stage, head at the right of frame, camera always on the same side. One book. Soldiers never look at or touch her. The 0-5s and 25-30s park must match exactly.
[Style] Photorealistic live-action. Natural skin, shallow depth of field, heavy film grain, volumetric atmosphere. Hair, cloth and dust react physically. No glossy CGI, no game rendering, no on-screen text.
The [Elaria] line describes our hair: burgundy roots, copper ends, green-hazel eyes. Change it to your own hair and eye color, so the words and the sheet agree.
Why this prompt works
She is the anchor. The prompt says her face, hair, clothes and pose are the only things that never change, and that she lies on her back the whole time. With one fixed point, the model is free to change everything else around her.
The change happens object by object. The prompt names each park object and what it becomes: the blanket a war banner, the pillow a bedroll, the cup a metal camp cup. That is why the transformation looks planned and you can follow it with your eyes.
Each stage starts with “She turns a page”. The page turn gives the model a clear moment to start the next change, so the story moves in steps that match the reading.
The camera stays low, at ground level, and passes a spear, legs and smoke in the foreground. That puts you inside the battle. A high camera would make the war look like a backdrop.
The end mirrors the start. The prompt asks for exactly the opening park, the same low view and one ash flake on the cover. So the reel loops well, and the ash tells the viewer it really happened.
Where it goes wrong
| What you got | The fix |
|---|---|
| A green patch stays around her while the rest changes | Keep the [Rule] block: “Nothing is fenced off; the ground she lies on changes too” |
| She sits up or stands | Keep “never sits up, never moves off her spot” |
| She wears the clothes from the sheet | Keep “Black sleeveless top, faded light-blue jeans, bare feet” in the [Elaria] line |
| A white studio wall shows up | Keep “Do not use the white studio background or the upright posture” |
| The dragons have four legs and wings | Keep the [Dragons] block: “wings as forelimbs, no front legs” |
| Soldiers stop and look at her | Keep “Soldiers never look at or touch her” |
| The park at the end is different from the start | Keep “The 0-5s and 25-30s park must match exactly” |
| It looks like a video game | Keep “No glossy CGI, no game rendering” and “heavy film grain” |
Questions people ask
Which AI model makes this video?
The video is Seedance 2.5, the images are GPT Image 2. We made both on Vosu. The video prompt is written for Seedance 2.5 at 30 seconds.
Can I put myself in it?
Yes. Make the character sheet from your own photo, add it as @Image1, and change the hair and eye colors in the [Elaria] line to yours.
Can I use a shorter length?
The prompt is timed for 30 seconds in five stages, and we only tested it at 30. For a shorter video, cut stages and fix the times, but keep the park at the start and the end.
Is it vertical?
Yes, 9:16, ready for Reels, TikTok and Shorts.
Is it free?
The prompts are free. Vosu is a paid service with a trial, so the images and the video use credits from your account.
Can I change the story in the book?
Yes. Keep the park, the [Rule] block and the object list, and rewrite the [Transformed world] block. We only tested the dragon war. For another story, give every park object a clear new form, the same way this prompt does.
Ready? Open Vosu, make your sheet and paste the video prompt.
Why this workflow runs here
Made for short-form
Trial available
One place for the video step
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
Cinematic Free Action Scene AI Video Prompt: Bike Jump and Fight
Action scene AI video prompt for Seedance 2.5. You jump a futuristic bike off a ramp, land in fire and fight on a burning highway. 15 seconds, one photo.
Cinematic Free Anime Borscht — The Whole Recipe in One 25-Second Video
One prompt cooks Ukrainian red borscht from raw meat to a served bowl, in glossy anime style. Six stages, hands and food only, vertical for Reels.
Cinematic Free APT AI Prompt — A Dog and a Cat Do the APT. Dance
Free APT AI prompt for Seedance 2.5. A dog sings and a cat plays drums in the APT. dance. Download the clip, make two sheets, swap in your own pets.