What you'll walk away able to do: film yourself saying one line, then drop yourself into a burning, falling, or lion-stalked danger scene that stops the scroll — using one reference image and one driving video, for a few dollars a hook.
This is the exact system behind the reel you just watched. Four steps. No green screen, no VFX skills.
The 4-step workflow
1. Record your real talking video. Shoot yourself on your phone saying the line for that hook, with the energy and hand mannerisms you want. This is your "driving video" — the AI keeps your exact words, timing, and gestures. Keep it simple: chest-up, decent light, look at the lens.
2. Generate your danger-scene image in Nano Banana Pro. Upload a photo of yourself and prompt the scene you want (hanging off a building, car speeding at you, lion behind you). This still is your "texture reference" — it tells the AI what the world and your look should be.
3. Combine the two in Higgsfield (Seedance 2.5). Load your talking video as the driving video and the Nano Banana image as the reference/texture. The prompt tells the scene how to move and escalate while you deliver your line. Out comes the finished hook: your real performance, inside the danger scene.
4. Roll the dice, then pick the winner. Generate 2–3 images and 2–3 videos per hook and keep the best one. AI video is a numbers game — the top pick out of a batch is what makes it look expensive.
The reusable prompt template
This is the skeleton for step 3. Swap the last two sentences for your scene:
Use @Video 1 as the driving video. Keep everything in the video the exact
same — exactly what the man says, with the exact same mannerisms. Have
@Image 1 as the texture reference. Full stop.
[WHAT THE SCENE IS DOING AS THE VIDEO STARTS].
The man in the video says exactly what he says with the exact same mannerisms,
and [WHAT HAPPENS ON THE LAST WORD].The magic is in the two brackets: set the danger in motion at the start, and land the payoff on the last word he speaks. That sync is what makes it feel directed instead of random.
3 done-for-you example prompts
Each maps to a spoken line from the reel.
Car barreling toward you
Line: "telling a story as a car is about to hit you from behind"
Use @Video 1 as the driving video. Keep everything in the video the exact same,
exactly what the man says with the exact same mannerisms. Have @Image 1 as the
texture reference. Full stop. As soon as the video starts the car is speeding
down the road, barreling toward the man. The man in the video says exactly what
he says with the exact same mannerisms, and as he finishes the last word the car
hits him and he flies out of scene.Hair catching fire
Line: the opening attention-grab
Use @Video 1 as the driving video. Keep everything in the video the exact same,
exactly what the man says with the exact same mannerisms. Have @Image 1 as the
texture reference. Full stop. As soon as the video starts there is no money in
the scene, and the lighter just starts to catch the hair on fire. The man in the
video says exactly what he says with the exact same mannerisms, and as he
finishes the last word half of the hair is on fire.Lion stalking from behind
Line: "you can be in the savanna with a lion approaching from behind"
Use @Video 1 as the driving video. Keep everything in the video the exact same,
exactly what the man says with the exact same mannerisms. Have @Image 1 as the
texture reference. Full stop. As soon as the video starts the lion is far in the
background and is starting to run toward the man. The man in the video says
exactly what he says with the exact same mannerisms as he turns around. After
saying the last word, the lion roars right in his face.Pro tips that separate good from viral
- Land the danger on the last word. Build tension across the line, pay it off on the final syllable, then cut. That's the beat that makes people rewatch.
- Batch and cherry-pick. 2–3 images x 2–3 videos per hook. Always.
- Match your image lighting to your video lighting so the composite reads real.
- Stack 4–5 hooks into one reel like the original: open on the most extreme one, escalate, end on a comment-keyword CTA.
Want this running for your brand instead of one-off? I build content + lead-gen systems that turn this into a repeatable pipeline. If you run a clinic or a personal brand and want your hooks (and the DMs they generate) on autopilot, DM me BUILD and I'll show you what it looks like.
🎨 Complete Prompt Library — all 6 hooks
Each hook has two prompts: the Nano Banana Pro image (your texture reference — generate it from a photo of yourself) and the Seedance 2.5 driving prompt (combine that image with your talking video). Every image prompt is written as an edit on a reference photo of you, so keep your input photo consistent — a plain black t-shirt works best.
1. Hair on fire
Nano Banana Pro image
Keep the exact same man from the reference image: same face, beard, hairstyle, expression, and black t-shirt. Preserve his identity and his intense wide-eyed stare into the camera.
He holds an orange lighter raised beside his head in one hand, already lit with a bright teardrop flame, the flame reaching the tips of his hair so a few strands are just beginning to catch and glow. No money in the scene. His free hand is open near his chest mid-gesture.
Relight the whole scene cinematically: a warm orange glow from the flame lighting one side of his face from above, deep moody shadows on the other side, a soft rim light separating him from the dark background. Cut any flat overhead lighting. Darken and blur the background into a clean, low-key cinematic backdrop with subtle warm bokeh.
Remove the small clip-on microphone from his shirt. Keep the watch and ring.
photorealistic, shot on 35mm, shallow depth of field, dramatic chiaroscuro lighting, ultra-detailed skin texture and beard, sharp catchlights in the eyes, realistic flame glow and rising smoke, cinematic color grade, 8k, hyper-realistic, no cartoon, no CGI lookSeedance 2.5 driving prompt
Use @Video 1 as the driving video. Keep everything in the video the exact same, exactly what the man says with the exact same mannerisms. Have @Image 1 as the texture reference. Full stop. As soon as the video starts there is no money in the scene, and the lighter just starts to catch the hair on fire. The man in the video says exactly what he says with the exact same mannerisms, and as he finishes the last word half of the hair is on fire.2. Interrogation room (bag reveal)
This one is a two-frame reveal: generate Frame A and Frame B, then load Frame A as the start frame and Frame B as the end frame in the video job.
Nano Banana Pro — Frame A (bag ON)
A man in a fitted black t-shirt sitting at a bare metal table in a dark concrete interrogation room, wrists handcuffed together in front of him resting on the table, a rough burlap sack pulled fully over his head. He leans slightly forward under a single harsh overhead spotlight that carves hard shadows, the rest of the room swallowed in black. Faint haze in the light beam, cold blue-grey tones against pooled shadow. Tense, ominous, captive stillness. Same camera angle and distance as a medium-close shot. photorealistic, shot on 35mm, shallow depth of field, cinematic color grade, hyper-realistic, 8kNano Banana Pro — Frame B (bag OFF, face revealed) — use your face reference at high strength
The same man from the reference sitting at the same bare metal table in the same dark concrete interrogation room, wrists handcuffed in front of him, a burlap sack caught mid-air just above and behind his head as an unseen hand yanks it away with motion blur. His hair is slightly disheveled, eyes wide then hardening into a cold defiant stare straight into the camera, jaw tight, mid-gasp. Same single harsh overhead spotlight, same haze, same cold blue-grey tones and pooled shadow, same camera angle and distance. Claustrophobic, high-stakes, thriller reveal. photorealistic, shot on 35mm, shallow depth of field, cinematic color grade, ultra-detailed skin texture, sharp catchlights in the eyes, hyper-realistic, 8kSeedance 2.5 motion (Frame A start, Frame B end)
A hand reaches in from the top of frame and abruptly rips the burlap sack off the man's head in one fast motion. His head jerks slightly with the pull, hair falls loose, and his expression snaps into a hard stare directly into the camera. The overhead light and room stay fixed, only the bag and his head move. Fast, violent yank, then he holds still. Handheld micro-shake, dramatic.3. Hanging off a building
Nano Banana Pro image
Keep the exact same man from the reference image: same face, beard, hairstyle, and black t-shirt. Preserve his identity and his intense wide-eyed stare toward the camera.
Place him clinging to the concrete ledge of a skyscraper, one hand gripping the edge with knuckles straining, his body dangling high above a city, wind pushing his hair and shirt. Dramatic downward angle so the street and traffic are tiny far below him.
Relight cinematically: cool overcast daylight, hard directional sun catching one side of his face, cold concrete greys, crisp shadows, a dizzying sense of height and depth. Sharp on him, atmospheric haze on the city below.
Remove the small clip-on microphone from his shirt. Keep the watch and ring.
photorealistic, shot on 35mm wide lens, shallow depth of field on the subject, vertigo-inducing perspective, ultra-detailed skin texture and beard, sharp catchlights in the eyes, cinematic color grade, 8k, hyper-realistic, no cartoon, no CGI lookSeedance 2.5 driving prompt
Use @Video 1 as the driving video. Keep everything in the video the exact same, exactly what the man says with the exact same mannerisms. Have @Image 1 as the texture reference. Full stop. As soon as the video starts one hand is hanging off the building. The man in the video says exactly what he says with the exact same mannerisms.4. Car in the street (suburban)
Nano Banana Pro image
Keep the exact same man from the reference image: same face, beard, hairstyle, and black t-shirt. Preserve his identity and his intense wide-eyed stare directly into the camera.
Place him standing in the middle of a quiet suburban residential street, facing the camera, arms slightly out mid-gesture. Directly behind him a car speeds toward him with headlights blazing, slightly motion-blurred to show speed. Line the street with suburban houses, driveways, parked cars, green lawns and trees.
Relight cinematically: warm late-afternoon golden-hour light with long shadows down the road, the car's headlights glowing behind him, a soft rim light on his shoulders separating him from the background. Sense of imminent impact.
Remove the small clip-on microphone from his shirt. Keep the watch and ring.
photorealistic, shot on 35mm, shallow depth of field, motion blur on the car, ultra-detailed skin texture and beard, sharp catchlights in the eyes, cinematic color grade, 8k, hyper-realistic, no cartoon, no CGI lookSeedance 2.5 driving prompt
Use @Video 1 as the driving video. Keep everything in the video the exact same, exactly what the man says with the exact same mannerisms. Have @Image 1 as the texture reference. Full stop. As soon as the video starts the car is speeding down the road, barreling toward the man. The man in the video says exactly what he says with the exact same mannerisms, and as he finishes the last word the car hits him and he flies out of scene.5. Burning money
Nano Banana Pro image
Keep the exact same man from the reference image: same face, beard, hairstyle, expression, and black t-shirt. Preserve his identity and his intense wide-eyed stare into the camera.
In his right hand, replace what he is holding with a fanned stack of crisp US one-hundred-dollar bills, Benjamin Franklin portraits clearly visible, held up toward the camera. In his left hand, he holds an orange lighter that is already lit, with a bright teardrop flame burning at the top, the flame positioned just below the edge of the cash as if about to set it alight.
Relight the whole scene cinematically: a warm orange glow from the flame illuminating his face and the bills from below, deep moody shadows filling the rest of the room, a soft rim light separating him from the dark background. Cut the flat overhead lighting. Darken and blur the background curtains and plant into a clean, low-key cinematic backdrop with subtle warm bokeh from string lights.
Remove the small clip-on microphone from his shirt. Keep the watch and ring.
photorealistic, shot on 35mm, shallow depth of field, dramatic chiaroscuro lighting, ultra-detailed skin texture and beard, sharp catchlights in the eyes, realistic flame glow and subtle smoke, crisp detail on the money, cinematic color grade, 8k, hyper-realistic, no cartoon, no CGI lookSeedance 2.5 driving prompt
Use @Video 1 as the driving video. Keep everything in the video the exact same, exactly what the man says with the exact same mannerisms. Have @Image 1 as the texture reference. Full stop. As soon as the video starts money starts to catch on fire. The man in the video says exactly what he says with the exact same mannerisms, and as he finishes the last word his finger starts to burn.6. Lion from behind
Nano Banana Pro image
Keep the exact same man from the reference image: same face, beard, hairstyle, and black t-shirt. Preserve his identity and his intense wide-eyed stare directly into the camera.
Place him in the foreground standing in tall golden savanna grass, facing the camera with a subtle knowing look, not turning around. Directly behind his shoulder a massive male lion with a full mane prowls in close, fully in focus, head level with his, backlit by a low sun.
Relight cinematically: warm golden-hour backlight rimming both the man and the lion's mane, floating dust in the air, acacia trees on the horizon, long shadows across the grass. Primal, silent-threat tension.
Remove the small clip-on microphone from his shirt. Keep the watch and ring.
photorealistic, shot on 35mm, shallow depth of field, warm golden-hour lighting, ultra-detailed skin texture and beard, detailed lion fur and mane, sharp catchlights in the eyes, cinematic color grade, 8k, hyper-realistic, no cartoon, no CGI lookSeedance 2.5 driving prompt
Use @Video 1 as the driving video. Keep everything in the video the exact same, exactly what the man says with the exact same mannerisms. Have @Image 1 as the texture reference. Full stop. As soon as the video starts the lion is far in the background and is starting to run toward the man. The man in the video says exactly what he says with the exact same mannerisms as he turns around. After saying the last word, the lion roars right in his face.