Turn Your Own Clips Into AI Visual Hooks

I add one impossible thing to plain phone clips of myself, keeping my real room

12 generations$2.91 total

Disclaimer: This is an independent creative test made with AI, for demonstration only. It isn't an ad for any brand shown or named, and isn't affiliated with, sponsored or endorsed by them. All trademarks belong to their owners

Let's build some visual hooks with AI and original footage.

The idea is simple. I shoot a quick clip of myself in my living room doing some random pose, then I use that real footage to build a hook out of it. I keep my face, my room and the way I actually moved, and I add one thing that couldn't possibly be there. I made three of these.

Here's my exact process.

First hook: feel the pressure

1. Start with a real clip of yourself

First I shoot a plain clip on my phone. For this one I sat on the floor with my back against a pillar and both arms up over my head, like I was holding something up. The pose is doing most of the work here, so I pick one that already looks like it means something.

2. Add one thing to a single frame

Then I take one frame from the clip where my face is clear and have the model redraw it with a single addition: a massive boulder resting on my hands. Same room, same angle, same clothes, same strained face. I keep the camera identical because the video has to start from this exact frame.

The same frame with a massive grey boulder resting on the man's raised hands, his face strained., made with GPT Image 2.6 Flare

Exact prompt

Keep the reference photo exactly as it is: the same room, the same camera angle, the same framing and crop, the same light, the same man in the same clothes and the same pose. Add only one thing: a massive boulder resting on his raised hands, which he is holding up above his head. THE VIBE: An ordinary living room, a guy sitting on the floor in sweatpants and slides, and he is straining to hold up a gigantic boulder that is pressing down on him. Everything else in the room is completely normal, which is what makes it funny. IDENTITY LOCK: Preserve the man exactly from the reference: short dark curly hair, thin-framed glasses, trimmed dark beard, warm medium skin, his black t-shirt, grey sweatpants and navy-and-white striped slides. Do NOT change his face, body or clothes. THE POSE (LOCKED): Exactly as in the reference: sitting on the tile floor with his knees bent, his back against the cream pillar, both arms raised with his elbows bent and his palms facing up beside and above his head, his face turned slightly up and to the side with a strained look. Do NOT alter it. THE SUBJECT: A gigantic, heavy, rough grey granite boulder, wider than his shoulders and much taller than his head, sits right on his upturned palms and fingertips. Its rough, cracked, chalky surface is pressed into his hands, its bulk filling the top of the frame and running past the top edge, so you only see its lower part. Small crumbs of rock and a little dust trickle down past his face and arms. His fingers are spread and pressed hard against the stone, the muscles in his forearms tense, his shoulders pushed up under the weight. THE FACE: Real, unretouched skin with fine texture and faint pores, mostly matte. No makeup. The same strained expression as in the reference. THE SETTING: Exactly the room from the reference, unchanged: the cream pillar behind him, the blue-and-gold patterned sofas, the rugs, the carved wooden cabinet and the speckled stone tile floor. THE LIGHT: Exactly the reference's light: soft warm daylight from the front, lighting him, the room and the underside of the boulder the same way. The boulder casts a heavy soft shadow down onto his head, shoulders and the pillar behind him, and its underside is darker than its lit front. THE KEY DETAILS: His fingers pressed into the rough stone, the falling crumbs and dust, the boulder's shadow on his head and shoulders, how much bigger the rock is than him, the totally normal room around him. THE MOOD: Deadpan absurd, a man under crushing pressure in his own living room. THE COMPOSITION: Vertical 9:16, framing identical to the reference, the man in the same place and at the same size, the boulder taking up the top third of the frame and continuing past the top edge. THE CAMERA: The same phone camera at the same slightly high, straight-on angle as the reference, deep focus, photoreal.

3. Start the video from the still and borrow your movement

Now the still becomes the first frame of the video, and my original clip goes in as a motion reference. I tell the model the room and the rock come from the still, and only my arm and head movements come from the clip. Then I ask for the one thing my clip can't give me: the rock shaking in my hands with dust falling on me.

Kling 4.0 $0.76

Exact prompt

Create a new video where a man in a black t-shirt, grey sweatpants and striped slides sits on the tile floor of an ordinary living room, back against a cream pillar, straining to hold a gigantic grey granite boulder up above his head with both hands. One continuous take, no cuts, locked camera. The boulder trembles and shakes in his hands under its own weight while he braces against it, crumbs and dust trickling down onto him. Everything about the room is completely normal, which makes it funny. Soft warm daylight. @image_1 is the starting frame: the room, the man, his clothes, the boulder and the camera all start exactly as in @image_1. @video_1 is the man's bracing performance and timing ONLY: take the small shifts of his arms and hands, the way he turns and tilts his head and glances around while he strains; ignore its framing and the fact that there is no rock in it. [SCENE 1] THE STARTING FRAME: exactly @image_1, the man sitting on the floor with his knees bent and his back against the pillar, both arms raised with his palms and spread fingers pressed up against the underside of the huge boulder, which fills the top of the frame and runs past the top edge, his face strained and turned slightly up to the side. Video animates FROM this position. THE SETTING (UNCHANGED FROM FRAME): the room, the sofas, the rug, the pillar, the cabinet and the tile floor stay completely still. The camera does not move. Only animate the man and the boulder: his arms and hands shift and re-grip under the weight, and his head turns and tilts as he strains, following the movement and timing of @video_1. The boulder shakes in his hands the whole time, small, heavy trembles and little jolts of a few centimetres side to side and up and down, as if it is barely being held, his forearms quivering with it. With each shake, small crumbs of rock and puffs of grey dust fall from the boulder onto his head, shoulders and the floor. The boulder always stays on his palms and never falls, and stays the same size and shape. The same soft warm daylight throughout, the boulder casting its heavy shadow on his head and shoulders. Photoreal. No text, no logos, no extra people, no warped hands, no extra fingers.

Here's my original clip next to the hook.

Combine clips $0.00

Exact prompt

Combine the clips in order

Second hook: the filler job gone wrong

1. Pick a clip where your face does the work

This time I filmed a close-up selfie with my hand on my chin, looking super serious and glancing around like I'm deep in thought. Nothing else happens in the clip. I kept it that simple on purpose, because I wanted my lips to be the only weird thing in the shot.

2. Change one feature on your face

I pulled one frame where I'm looking straight at the camera and asked for that exact photo back with only my lips changed, blown up like an extreme lip filler job that went way too far. Same angle, same hand on my chin. The video is going to follow my real clip frame by frame, so the camera can't move even a little.

The same selfie frame with the man's lips hugely swollen like an extreme lip filler job, his face still deadpan., made with GPT Image 2.6 Flare

Exact prompt

Keep the reference photo exactly as it is and change only the man's lips. Same camera angle, same low selfie framing, same crop, same ornate cream room with the chandelier and gold curtains, same light, same black top, same hand resting on his chin. Preserve the man exactly from the reference: same face, eyes, brows, hair, beard and skin. THE VIBE: A deadpan, serious, thoughtful selfie of a man whose lips have been catastrophically overfilled, the most extreme, over-the-top botched lip-filler job imaginable, a cartoonish trout pout. The joke is that his eyes and expression stay completely serious while his lips are gigantic. IDENTITY LOCK: The man with short dark curly hair, a trimmed dark beard with grey in it, and warm medium skin, exactly from the reference. Only his lips change. THE SUBJECT: His own human lips, absurdly, extremely overfilled to about five times their normal size, so swollen they dominate the lower half of his face. Each lip is a huge, tight, rounded balloon, wider than his nostrils on both sides and almost reaching the corners of his jaw. The upper lip is so overfilled it flips fully up and outward, its edge curling up to touch the base of his nose and hiding his moustache. The lower lip is an enormous swollen sausage that bulges far out and hangs down heavily over his chin, its underside resting on the top of his fingers. Together they push forward several centimetres from his face in a giant pursed trout pout. The lip skin is stretched shiny and drum-tight from the filler, still real lip skin with faint vertical creases, a natural pink-red lip colour and a slightly darker overdrawn lip line. They are unmistakably human lips, not a beak, a bill or a prop, just wildly, hilariously overdone. The beard grows right up to their edges and they are lit by the same warm room light. His mouth is closed. His eyes look straight at the lens, calm and unimpressed. THE FACE: Real, unretouched skin with fine visible texture and faint pores, mostly matte. No makeup. Beard as individual hairs. THE SETTING: Unchanged from the reference: the ornate carved plaster ceiling, the lit crystal chandelier, the gold draped curtains and the cream pillar behind him. THE LIGHT: Unchanged from the reference: KEY is soft warm daylight from the front and above falling on his face, with a bright highlight on the drum-tight swollen lips. FIELD is the warm bright cream room, about as bright as his face. Warm natural grade. THE KEY DETAILS: The upper lip curled up against the base of his nose, the giant lower lip hanging over his chin and resting on his fingers, the stretched shiny lip skin, the lips pushing far out from the face and casting a soft shadow onto his beard and hand, his completely serious eyes. THE MOOD: Deadpan absurd comedy. THE COMPOSITION: Vertical 9:16, framing identical to the reference, his face in the same place, head about 37% down from the top, the hand on his chin at the bottom. THE CAMERA: The same phone front-camera selfie from slightly below, exactly as in the reference.

3. Swap the still back into the clip

This one works a little differently. Nothing new has to move, so instead of starting from the still, I use my clip as the base and the still as the reference. I told the model to keep every glance, every head turn and the camera exactly the same, and only take the lips from the still. So every look you see in the final clip is one I actually did.

Kling 4.0 $0.94

Exact prompt

Create a new video from @video_1: the same man in the same ornate cream room, doing exactly the same thing, a deadpan, thoughtful selfie with his hand on his chin while he glances around, except his lips are now the enormous, extremely overfilled filler lips from @image_1. One continuous take, no cuts. Replace the man in @video_1 with the man in @image_1, who is the same man with giant ballooned lips. Keep his face, eyes, beard, hair and skin, and take the huge, drum-tight, shiny pink-red overfilled lips from @image_1 exactly. Keep all of his motion, timing, glances, head movement and hand movement, and the camera, exactly the same as @video_1. Change nothing else. [SCENE 1] The same low front-camera selfie as @video_1: the man sits under the lit crystal chandelier with the gold curtains and cream pillar behind him, his hand resting on his chin, his giant overfilled lips from @image_1 pushing far out over his hand. He moves exactly as in @video_1, looking at the lens, then glancing up and to the side, then back, with the same small head turns and the same hand on his chin. The huge swollen lips move with his face, staying the same size and shape the whole time, with a slight soft jiggle as his head moves. His eyes and expression stay completely serious throughout. The same soft warm daylight from the front lights his face and the shiny lips, and the room stays as bright and warm as in @video_1. No warping of his face, hands or lips, the lips never shrink back to normal, no extra fingers, no text, no logos.

And here they are next to each other. I'm glancing around the exact same way in both, only my lips changed.

Combine clips $0.00

Exact prompt

Combine the clips in order

Third hook: the meditation cloud

1. Shoot something calm

For the last one I sat cross-legged on the floor, eyes closed, arms out, meditating. Super calm and nothing happening, which is exactly what I wanted for this one.

2. Put the cloud under you

I grabbed a frame and asked for the same shot with me sitting on a small fluffy cloud, floating about knee height above the tiles instead of sitting on the floor. Everything else in the room stays totally normal. I also asked for a soft shadow on the tiles under the cloud so it actually looks like it's hovering.

The same frame with the man meditating on a small fluffy cloud hovering above the tile floor., made with GPT Image 2.6 Flare

Exact prompt

Keep the reference photo exactly as it is: the same room, the same camera angle, the same framing and crop, the same light. Change only one thing: the man is now sitting cross-legged on a small fluffy cloud that hovers just above the floor, instead of sitting on the floor. THE VIBE: A calm, serious meditation in an ordinary living room, except he is floating on his own little cloud. Everything else about the room is completely normal, which is what makes it funny. IDENTITY LOCK: Preserve the man exactly from the reference: short dark curly hair, thin-framed glasses, trimmed dark beard, warm medium skin, his black t-shirt, grey sweatpants and navy-and-white striped slides. Do NOT change his face, body or clothes. THE POSE (LOCKED): Exactly as in the reference: sitting cross-legged, back upright, eyes fully closed, both arms held out to the sides with his fingers in a relaxed meditation gesture. Do NOT alter it. THE SUBJECT: Under him, in place of the floor he was sitting on, is a single small, dense, puffy white cumulus cloud, just big enough to sit on, its soft rounded top sinking slightly under his weight around his crossed legs and slides. The cloud is a natural, realistic white cloud with soft billowing edges and a few wisps curling off its sides, not a cartoon and not yellow. The cloud and the man hover about knee height above the tiles, with clear empty space between the bottom of the cloud and the floor. THE FACE: Real, unretouched skin with fine texture and faint pores, mostly matte. No makeup. Eyes fully closed, a calm, peaceful expression. THE SETTING: Exactly the room from the reference, unchanged: the cream walls and pillar, the gold patterned curtains, the blue-and-gold patterned floor sofas and cushions, the rugs, and the speckled stone tile floor with its dark grout lines. THE LIGHT: Exactly the reference's light: soft, warm indoor daylight from the front and above, lighting him and the cloud the same way it lights the room. The cloud's top is softly lit and its underside a little shaded, and it casts a soft, diffuse shadow onto the tiles directly below it. THE KEY DETAILS: The gap between the cloud and the floor, the soft shadow on the tiles under the cloud, the cloud top denting under his legs, his closed eyes, the totally normal room around him. THE MOOD: Deadpan, serene, quietly absurd. THE COMPOSITION: Vertical 9:16, framing identical to the reference, the man in the same place and at the same size, raised slightly higher by the hovering cloud, with the tile floor still filling the lower part of the frame. THE CAMERA: The same phone camera at the same low, straight-on angle as the reference, deep focus, photoreal.

3. Make the cloud move with you

Same setup as the first hook: the cloud still is the first frame and my clip drives my arms. The new part is the cloud itself. I asked for it to rock gently from side to side in time with my arms, like a slow cradle, and it picked up the rhythm of my real movement.

Kling 4.0 $0.94

Exact prompt

Create a new video where a calm man in a black t-shirt, grey sweatpants and striped slides meditates cross-legged on a small fluffy white cloud that hovers just above the tile floor of an ordinary living room with patterned sofas and gold curtains. One continuous take, no cuts, locked camera. He stays serene with his eyes closed while the cloud gently rocks back and forth under him in rhythm with the slow movement of his arms. Everything about the room is completely normal, which makes it quietly funny. Warm, soft indoor daylight. @image_1 is the starting frame: the room, the man, his clothes, the cloud and the camera all start exactly as in @image_1. @video_1 is the man's meditation performance and timing ONLY: take the slow, flowing movement of his arms and hands and the small movements of his head and shoulders; ignore its floor, the fact that he sits on the floor, and its framing. [SCENE 1] THE STARTING FRAME: exactly @image_1, the man sitting cross-legged on the hovering cloud about knee height above the tiles, both arms held out to the sides with his fingers in a relaxed meditation gesture, eyes closed, the patterned sofas, curtains and pillar behind him. Video animates FROM this position. THE SETTING (UNCHANGED FROM FRAME): the room, the sofas, the curtains, the pillar, the rugs and the tile floor stay completely still. The camera does not move. Only animate the man and the cloud: his arms drift slowly and smoothly up and out and back in, his hands floating through the air, following the movement and timing of @video_1. He stays upright, eyes closed, calm. The cloud rocks gently from side to side like a slow cradle, in time with his arms: as his arms sweep one way the cloud tilts a little to that side, and as they sweep back it tilts back the other way, a soft, slow sway of only a few degrees, with the man rocking along with it. The cloud stays at the same height above the floor while it rocks, its soft edges churning and a few wisps curling off its sides. The soft shadow on the tiles under the cloud shifts slightly side to side with the rocking. The gap between the cloud and the floor stays visible the whole time; he never touches the floor. The same soft, warm indoor daylight throughout. Photoreal. No text, no logos, no extra people, no warped hands, the cloud stays the same size and stays white.

And the original next to it.

Combine clips $0.00

Exact prompt

Combine the clips in order

Why it works

The reason this works is simple. I'm not asking the AI to invent me from nothing. I hand it real footage of me moving in my real room and only let it add the one impossible thing. On the first and third hooks that thing also had to move, so the still became the first frame and my clip handled my movement. On the second nothing new had to move, so my clip stayed the base. Either way every movement is really mine, so the clips hold up even though none of it ever happened.

Your turn: grab one clip of yourself doing a simple pose and pick one impossible thing to add. That's step one.

If you'd rather skip the prompting, Emero writes the prompts for you and keeps your look and your clips ready for the next hook, all at cost.

Total for this workflow$2.91

Get one of these every day

I test new workflows daily and send them with the exact prompts, the models and the costs

6,526 generations tested so far

More workflows

See all