Kling 4.0 + GPT Image 2.6 Flare - how to make AI videos of yourself in six costumes
Six phone clips, six costumes, one script, with my real voice and motion kept
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
I wanted to make a reel where every line I say comes with a brand new outfit. So I filmed six short clips of me talking in the same hallway and turned each one into a different character: a pizza delivery guy, a TV painter, a hype man with a giant clock necklace, a robot, a builder and a professor. Every clip keeps my real voice, my real moves and my real room.
Here's my exact process.
1. Write the script, then record it clip by clip
First, I write what I want to say and split it into short lines. Each line becomes its own clip, about 3 to 5 seconds long, with one action in it. I film them all in the same spot so the room matches from clip to clip.
Short clips matter more than you'd think. The AI edit puts one look on the whole clip, so one line gets one costume. You also pay for every second the model processes, so a 7 second clip costs about double a 3 second one. My professor clip ran almost 7 seconds and it was the priciest of the six.
Quick warning: iPhones film in HDR and AI makes standard video. Mix the two in your editor and the colors won't match. Turn HDR off before you film, or use the converted copy of your clip instead of the one in your camera roll.
2. Have Claude watch and listen to each clip
Then I send each clip to Claude and have it watch the whole thing and transcribe what I say, word by word. That tells me what my hands are actually doing and which frame to build the look from.
For the frame I want my face clear and looking at the camera. In the robot clip I lean down and look at the floor halfway through, so I used a frame from later on, when I'm facing the lens again.
3. Give your hands a job
This is the step I'd tell anyone to slow down on. Look at what your hands are doing in the clip and build the costume around it.
In the painter clip I'm making painting strokes with my left hand, so the brush goes in that hand, with the canvas right in front of me so every stroke lands on it. In the builder clip I gesture out to the side, so the giant wrench goes in that hand. In the robot clip I'm literally doing the robot, so my whole body becomes the machine. And in the pizza clip my palm is flat, which is the perfect spot for a tower of twelve pizza boxes.
The AI keeps your movement exactly as you filmed it, so if the prop fights your real motion, the video looks off.
4. Build the still from your clip
Now I have Claude pull that frame and edit it into the costume with an image model. The prompt starts by locking everything I want to keep: same camera angle, same framing, same room, same light, same pose. Then it changes only the outfit and the prop.
Two things I always put in: "preserve his face exactly, no aging", and a real description of every piece of the costume. For the hype man look I had Claude research the real signature outfit first. That way I got the horned helmet, the blacked-out shades, the rhinestone suit, the gold chains and the giant clock instead of a generic guess.
5. Send it to the editing model
Last, the still and the original clip go to a video editing model. The clip is the base, so my motion, timing and lip sync stay mine. The still tells it what I should look like.
The video prompt does two jobs. It says "lips, jaw and speech timing exactly as in the original, do not alter the mouth shapes", and it spells out the physics of anything new. The pizza tower sways and lags behind my hand like I'm barely balancing it. The clock bounces against my chest when I step. The wrench swings with real weight. When I've left those lines out, I've had props come out stiff, and once even lava that splashed like water.
One thing to know: text in AI video is still hit or miss. My chalkboard has the five steps of this process written on it, and one word came out misspelled. If the text has to be perfect, add it in your editor.
All six, still and video
Pizza delivery guy. The tower of boxes sits on my flat palm and sways as my arm moves.

Exact prompt
Keep the first reference photo exactly as it is: same camera angle, framing, crop, room, light, pose and expression. Change only two things: his outfit, and the object on his flat right palm. Photorealistic, shot natively in camera, not a composite. THE VIBE: Deadpan comedy. A man explaining something to camera while dressed as the most over-the-top pizza shop worker imaginable, balancing a ridiculous tower of pizza boxes like it's totally normal. IDENTITY LOCK: The man from the first reference. Preserve his face exactly as in the reference, no aging, same hair. THE SUBJECT: Same pose as the first reference: standing, his right arm extended out to the side at chest height with the palm flat and facing up, his left hand raised holding the small grey clip-on microphone near his chin. He now wears an over-the-top pizzeria uniform: a loud red and white checkered short-sleeve work shirt with a big round embroidered pizza patch on the chest, a long red apron dusted with white flour and a few tomato sauce smudges, a white paper pizza-chef neckerchief, and on his head an oversized bright red pizzeria cap with a huge soft foam pizza slice standing upright on top of it, cheese and pepperoni included. THE KEY DETAILS: The folded paper on his flat right palm is gone. In its place, balanced on that same flat palm, is a towering stack of exactly twelve square white cardboard pizza boxes with simple generic red line-art on the lids and no brand names, stacked straight but slightly leaning, rising from his palm to well above the top of his head, a few lids slightly lifted with steam curling out. His palm clearly supports the bottom box; the stack is big and dominant in the frame. His left hand still holds the microphone exactly as in the reference. THE SETTING: The same hallway as the first reference, unchanged: the tall polished wooden double doors, the carved plaster band above, the tile floor. THE LIGHT: KEY: the same warm overhead indoor light as the first reference, soft and slightly from above, on his face, the uniform and the boxes, with matching soft shadows under each box edge. FIELD: the wooden doors at the same brightness as the reference. Warm, natural, standard dynamic range. THE MOOD: Absurd, deadpan, funny. THE COMPOSITION: Vertical 9:16, same framing as the first reference. The pizza box tower rises from his palm on the left side of the frame and runs up past the top edge of the frame. Him centered as in the reference. THE CAMERA: iPhone, handheld, natural, exactly the capture of the first reference.
Exact prompt
Create a new video that takes @video_1, the man talking to camera in a hallway, and dresses him exactly as in @image_1, keeping his real motion, camera, timing and lip sync. One continuous ~3-second talking shot. [SCENE 1] THE STARTING LOOK: exactly as @image_1. He wears the over-the-top pizzeria uniform: the red and white checkered shirt with the round pizza patch, the long flour-dusted red apron with sauce smudges, the white neckerchief, and the oversized red cap with the big foam pepperoni slice standing on top. On his flat right palm sits the tall tower of twelve white pizza boxes from @image_1, at full size, rising above his head; the folded paper from @video_1 is gone. His left hand holds the small grey microphone at his chin exactly as in @video_1. His face is preserved exactly, no aging. The hallway, the wooden doors and the warm light are unchanged. He is talking: lips, jaw and speech timing exactly as in @video_1, do not alter the mouth shapes. The pizza box tower is locked to his right palm and moves with his real arm. As his arm shifts, the whole tall stack sways and teeters, the upper boxes leaning a little further than the bottom ones and lagging behind his hand, then rocking back, like a heavy tall stack he is barely balancing. The boxes stay stacked and never fall or vanish. Faint steam drifts up from the top box. The foam pizza slice on his cap bobs slightly as his head moves. Photoreal, his face locked, like real props.
TV painter. The brush follows my real strokes, and a small evergreen tree builds on the canvas.

Exact prompt
Keep the first reference photo exactly as it is: the same camera angle, the same framing and crop (do not zoom out, do not widen the shot), the same room and light, and his exact pose and side-on stance at the right edge of the frame. Change only three things: his hair, his outfit, and add a paintbrush and a canvas he is painting on. Photorealistic, shot natively in camera, not a composite. THE VIBE: Wholesome, gentle TV-painter energy, played completely straight. A calm man mid-stroke, dabbing a happy little tree onto his canvas. IDENTITY LOCK: The man from the first reference, seen side-on from behind his right shoulder. Preserve his face, beard and profile exactly as in the reference, no aging. THE SUBJECT: Exactly the pose of the first reference: standing side-on at the right edge of the frame, facing left, his right hand raised to his mouth holding the small grey microphone, his left arm bent with the left hand held out just in front of his chest at shoulder height. His left hand now holds a long wooden fan paintbrush loaded with green paint, held like a painter mid-stroke. His hair is now a HUGE, perfectly round, tightly curled permed afro, light brown, at least twice the width of his head, a big soft spherical halo of springy curls framing his head on all sides. He wears a light blue denim button-up shirt with the sleeves rolled to the elbows, and blue jeans. THE KEY DETAILS: A large stretched canvas on a wooden studio easel stands DIRECTLY in front of him, close, at arm's reach, its painted surface exactly where his left hand is, so the tip of the brush is touching the canvas. The canvas faces him and is angled so its painted face is readable from the camera at a three-quarter angle. On it, a wet-on-wet oil landscape: tall snow-capped mountains, soft blue sky with fluffy clouds, a row of dark evergreen trees, and a calm lake reflecting it, with one fresh green tree being painted right where the brush touches. A kidney-shaped wooden palette with blobs of white, blue, green, brown and yellow paint rests on the easel's ledge below the canvas. THE SETTING: The same hallway as the first reference, unchanged: the tall polished wooden doors behind, the carved plaster band above, the tile floor. THE LIGHT: The same warm overhead indoor light as the first reference, soft, on his hair, shirt, the brush and the canvas, with matching soft shadows. Warm, natural, standard dynamic range. THE COMPOSITION: Vertical 9:16, IDENTICAL framing to the first reference. He stays at the right edge, side-on, at the same size in frame. The canvas and easel fill the space directly in front of him on the left, close to his hand. THE CAMERA: iPhone, handheld, natural, exactly the capture of the first reference.
Exact prompt
Create a new video that takes @video_1, the man standing side-on in a hallway talking into a microphone while making painting strokes with his left hand, and turns him into the TV-style painter from @image_1, keeping his real motion, camera, timing and lip sync. One continuous ~4.5-second shot. [SCENE 1] THE STARTING LOOK: exactly as @image_1. He has the big round curly permed afro, the light blue denim shirt with rolled sleeves and jeans. His right hand holds the small grey microphone at his mouth exactly as in @video_1. His left hand holds the long wooden fan paintbrush from @image_1. Directly in front of him stands the wooden easel with the large landscape canvas from @image_1 (snow-capped mountains, evergreens, a calm lake), close at arm's reach, with the palette resting on the easel ledge. His face is preserved exactly, no aging. The hallway, wooden doors and warm light are unchanged. He is talking: lips, jaw and speech timing exactly as in @video_1, do not alter the mouth shapes. The paintbrush is locked in his left hand and follows his real left-hand strokes from @video_1 exactly. Each time his hand moves forward, the brush tip presses onto the canvas surface and dabs on fresh dark green paint, so a small evergreen tree visibly grows on the canvas stroke by stroke; by the last frame the new little tree is clearly finished. The paint stays on the canvas and never disappears. The easel and canvas stay planted and still; only the brush and the new paint move. His big curls bounce softly as his head moves. Photoreal, like a real painter at a real easel.
Hype man. A horned helmet, blacked-out shades, a rhinestone suit and the giant clock on a gold chain.

Exact prompt
Keep the first reference photo exactly as it is: the same camera angle, framing and crop (do not zoom in or out), the same room and light, and his exact pose and expression. Change only his outfit and accessories. Photorealistic, shot natively in camera, not a composite. THE VIBE: Loud, flashy, over-the-top old-school hip-hop hype-man energy, worn with total confidence. Every accessory is absurdly oversized and sparkly on purpose. IDENTITY LOCK: The man from the first reference. Preserve his face exactly as in the reference, no aging. His mouth and teeth stay exactly as they are. THE SUBJECT: Exactly his pose from the first reference: standing full-body facing the camera, his right hand raised to his chin holding the small grey microphone, his left arm at his side. He now wears: - a horned Viking helmet: a rounded silver metal helmet with two big curved cream-colored horns sticking out the sides; - ENORMOUS oversized sunglasses with thick black frames and fully blacked-out dark lenses, much wider than his face; - an enormous round analog clock, about the size of a large dinner plate, its rim fully encrusted with sparkling clear crystals, a white face with big black numerals and black hands, hanging from a thick chunky gold chain so it rests in the middle of his chest and covers most of his torso; - a flashy custom suit: a bright red jacket and matching trousers completely covered in sparkling rhinestones that catch the light, over a black shirt; - extra thick gold chains layered around his neck, and gold rings on his fingers; - clean white sneakers. THE SETTING: The same hallway as the first reference, unchanged: the tall polished wooden doors, the carved plaster band above, the tile floor. THE LIGHT: The same warm overhead indoor light as the first reference, glinting off the rhinestones, the crystal clock rim, the gold chains, the helmet and the glossy lenses, with matching soft shadows. Warm, natural, standard dynamic range. THE COMPOSITION: Vertical 9:16, IDENTICAL framing to the first reference, full body visible. The giant crystal clock is the dominant thing on his chest. THE CAMERA: iPhone, handheld, natural, exactly the capture of the first reference.
Exact prompt
Create a new video that takes @video_1, the man stepping forward in a hallway and talking into a microphone, and dresses him exactly as in @image_1, keeping his real motion, camera, timing and lip sync. One continuous ~3.8-second shot. [SCENE 1] THE STARTING LOOK: exactly as @image_1. He wears the horned silver Viking helmet with the two big curved cream horns, the oversized blacked-out sunglasses, the red rhinestone-covered suit over a black shirt, layered thick gold chains, gold rings, white sneakers, and the giant dinner-plate-sized clock with the crystal-encrusted rim hanging on a thick gold chain in the middle of his chest. His right hand holds the small grey microphone at his chin exactly as in @video_1. His face is preserved exactly, no aging. The hallway, wooden doors and warm light are unchanged. He is talking: lips, jaw and speech timing exactly as in @video_1, do not alter the mouth shapes. He steps forward and gestures exactly as in @video_1. The giant clock is a heavy object on its chain: it swings and bounces against his chest as he steps, lags a little behind his body, then settles, and it stays on his chest the whole time. The rhinestones on the suit and the crystals on the clock rim glint and sparkle as he moves through the light. The helmet and sunglasses stay firmly on his head. Photoreal, like a real flashy costume on a real person.
Robot. A full chrome body with a glowing chest core that hits every robot-dance pose.

Exact prompt
Keep the reference photo exactly as it is: same camera angle, framing, crop, room, light, pose and expression. Change only one thing: his whole body below the face is now a full, cinematic, movie-quality humanoid robot, as if the first reference were a frame from a big-budget sci-fi film with flawless VFX. THE VIBE: An over-the-top, fully committed robot transformation, a man mid robot-dance who has literally become a chrome machine. It should look like real practical-plus-CG creature work, heavy, metallic and believable, never a costume, never a filter, never a sticker pasted on top of the photo. IDENTITY LOCK: The man from the first reference. Preserve his face exactly as in the reference, no aging. His real eyes, eyebrows, nose, mouth, lips and jaw stay fully human and fully visible, unchanged, in the same expression. THE SUBJECT: Everything except his face is robot. His head is encased in a polished chrome robot helmet that wraps his skull, forehead edge, ears and neck, with the human face showing through a clean open face-plate opening, the metal meeting his skin in a tight machined seam with tiny rivets, so his face reads as set into the machine. His neck becomes stacked chrome rings and exposed hydraulic pistons. His torso is a brushed-steel chest plate with panel seams, a small glowing cyan power core in the centre of the chest, and vent slats on the sides. His arms are segmented chrome with visible ball-joint elbows and shoulders, black rubber gaskets in the joints, and thin bundles of braided cable running along the forearms. His hands are articulated robot hands with jointed steel fingers, holding the exact same pose as in the first reference: his right hand, on the left of frame, raised near his chest pinching the same small dark object between metal fingertips; his left hand, on the right of frame, raised flat with fingers spread, palm angled, a stiff robot-dance pose. His legs are chrome-and-graphite robotic legs with knee pistons and armoured shin plates, standing in the same stance. THE KEY DETAILS: Real surface wear on the metal: fine scratches, slight scuffs on the edges, faint fingerprint smudges, small hex bolts at every plate corner, a few amber status LEDs on the forearms. Joints show depth: dark gaps between plates, visible inner mechanics at the elbows and knees. The cyan chest core casts a faint cyan glow onto the chest plates and the underside of his chin. THE SETTING: Same room exactly as the first reference: the tall polished wooden double doors behind him, the carved white plaster frieze above, the cream tile floor and the red mat at bottom left. Nothing in the room changes. THE LIGHT: KEY: the same warm overhead indoor light as the first reference, soft, from above and slightly to frame left, now reflecting hard specular highlights off the chrome, with the warm wood doors and cream walls reflected and bent in the curved metal surfaces. FIELD: the room stays at the same brightness as the reference. The metal is lit BY the room, sharing its warm cast, with real contact shadows under the arms and between plates. Colour grade unchanged from the reference. THE COMPOSITION: Same framing as the first reference, full body standing centre-right, head in the upper third, nothing cropped differently. THE CAMERA: Shot on an iPhone, same lens, same distance and same slight angle as the first reference, natural phone sharpness and noise.
Exact prompt
Create a new video where the man in @video_1 does his exact robot dance and talks to camera in the same room, but his whole body is now the cinematic chrome humanoid robot from @image_1: polished chrome helmet framing his real human face, stacked chrome neck rings with hydraulic pistons, a brushed-steel chest plate with a glowing cyan power core, segmented chrome arms with ball-joint elbows, braided cables and amber forearm LEDs, jointed steel robot fingers, and armoured chrome-and-graphite legs with knee pistons. It looks like big-budget sci-fi VFX, heavy and metallic, never a costume or a filter. His motion, timing, framing and camera stay exactly as in @video_1. [SCENE 1] The robot body from @image_1 is locked to his real body from the first frame and follows every move of @video_1 exactly: each stiff stop-start robot-dance hit, the hand pinched near his chest, the flat open hand, the lean down and the rise back up. The metal moves like real heavy machinery: the elbow and shoulder ball joints rotate visibly, the plates slide over each other at every bend, the neck rings compress and extend as his head moves, and every robot-dance hit stops dead with a tiny mechanical settle in the plates, as if servos lock in place. The cyan chest core pulses softly the whole time and casts a faint cyan glow on the chest plates and under his chin. The warm overhead room light slides across the chrome as he moves, with the wooden doors and cream walls reflected and bending in the curved metal. The robot parts stay on him in every frame and never flicker, melt or disappear. His face stays fully human and preserved exactly as in @video_1, no aging. Lips, jaw and speech timing exactly as in @video_1, do not alter the mouth shapes. The room stays unchanged: the tall wooden double doors, the carved white plaster frieze above, the cream tile floor and the red mat at bottom left. Shot on an iPhone, same lens, same distance, same handheld camera as @video_1, original audio kept.
Builder. A hard hat, a plaid shirt, overalls, a loaded tool belt and a wrench longer than my forearm.

Exact prompt
Keep the reference photo exactly as it is: same camera angle, framing, crop, room, light, pose and expression. Change only one thing: he is now dressed in a wildly over-the-top cartoon-style construction builder outfit and grips a gigantic wrench in his outstretched hand. THE VIBE: A cheerful, absurdly exaggerated builder look, as if a kids'-show construction worker came to life at real-world scale. Everything is bigger, brighter and more stuffed than real work gear, but it is real fabric, real plastic and real steel, photographed for real in this room, never a filter or a sticker. IDENTITY LOCK: The man from the first reference. Preserve his face exactly as in the reference, no aging. His eyes, mouth and expression stay unchanged and fully visible. THE SUBJECT: On his head sits an oversized glossy bright-yellow plastic hard hat, about twice normal size, with a wide brim, tilted slightly back. He wears a red-and-orange plaid flannel shirt with the sleeves rolled to the elbows, under bright denim-blue bib overalls with big brass buckles on the straps and a chunky front bib pocket stuffed with a folding ruler and a carpenter's pencil. Around his waist is a huge brown leather tool belt overloaded with tools: a claw hammer, screwdrivers, pliers, a tape measure and a spirit level all hanging off it. On his feet are chunky tan work boots with thick soles. Same pose as the first reference: his left hand, on the right of frame, still raised to his chin in the same thinking gesture; his right hand, on the left of frame, out to the side at the same height, now closed in a firm grip around the handle of the giant wrench. THE KEY DETAILS: The wrench is enormous, a chrome-steel adjustable wrench about as long as his whole arm, with a thick handle and a big open jaw. He holds it by the handle end, so the head of the wrench extends outward and slightly down toward the lower left of frame, the jaw clearly visible. It has real weight: brushed metal, a few scuffs, a knurled adjusting screw. The hard hat, overalls and belt show light real-world wear: small scuffs and dust. THE SETTING: Same room exactly as the first reference: the tall polished wooden double doors behind him with the brass handle, the carved white plaster frieze above, the cream tile floor and the red mat at bottom left. Nothing in the room changes. THE LIGHT: KEY: the same warm overhead indoor light as the first reference, soft, from above, putting a bright specular highlight on the top of the glossy yellow hard hat and a long highlight down the chrome wrench, with soft shadow under the hat brim on his forehead. FIELD: the room stays at the same brightness as the reference. Clothing colours read true: bright yellow hat, red-orange plaid, mid-blue denim, warmed slightly by the room light. Real contact shadows from the tool belt and wrench. Colour grade unchanged. THE COMPOSITION: Same framing as the first reference, full body standing centre-right, the wrench reaching into the open space on the left of frame. THE CAMERA: Shot on an iPhone, same lens, same distance and same slight angle as the first reference, natural phone sharpness and noise.
Exact prompt
Create a new video where the man in @video_1 talks to camera in the same room with the exact same gestures, but he is now dressed in the over-the-top builder outfit from @image_1: a glossy bright-yellow hard hat, a red-orange plaid flannel shirt with rolled sleeves, dusty denim-blue bib overalls with brass buckles and a stuffed bib pocket, a huge brown leather tool belt loaded with a hammer, pliers, screwdrivers, a tape measure and a yellow spirit level, and tan work boots. His outstretched hand grips the giant chrome adjustable wrench from @image_1. His motion, timing, framing and camera stay exactly as in @video_1. [SCENE 1] The outfit and wrench from @image_1 are on him from the first frame and follow every move of @video_1 exactly. His hand stays at his chin in the same thinking gesture. His other hand makes the same gestures out to the side as in @video_1, but now it grips the handle of the giant wrench from @image_1 the whole time, at full size, about as long as his arm; the wrench is locked to his hand and moves with it. The wrench is heavy solid steel: it swings with real weight as his hand gestures, its head lagging a moment behind his wrist and dipping slightly at the end of each gesture before settling, never floating or bending. The tools on the belt sway and clink gently as his body shifts, the hammer swinging a little on its loop, and the hard hat stays firmly on his head. The warm overhead room light slides along the chrome wrench and the glossy yellow hat as he moves. Nothing new appears or disappears. His face stays preserved exactly as in @video_1, no aging. Lips, jaw and speech timing exactly as in @video_1, do not alter the mouth shapes. The room stays unchanged: the tall wooden double doors with the brass handle, the carved white plaster frieze above, the cream tile floor and the red mat at bottom left. Shot on an iPhone, same lens, same distance, same handheld camera as @video_1, original audio kept.
Professor. Tweed with elbow patches, a bow tie, an oak lectern and the five steps on the chalkboard.

Exact prompt
The first reference is the man's identity: preserve his face exactly as in the first reference, no aging, no glasses. The second reference is the latest version of the image: keep it exactly as it is (same camera angle, framing, crop, room, light, pose, expression, professor outfit, lectern and everything on it). Change only one thing: the writing on the chalkboard. THE VIBE: The same warm old-school lecture moment as the second reference, but the chalkboard now teaches his actual lesson, hand-written in chalk by a professor mid-lecture. IDENTITY LOCK: The man from the first reference. Preserve his face exactly, no aging. His eyes, mouth and expression stay unchanged. THE SUBJECT: Same as the second reference: brown herringbone tweed blazer with the suede elbow patch on his raised elbow, burgundy sweater vest, pale blue oxford shirt, green striped bow tie, one hand raised at his chin, the other resting on the lectern edge. THE KEY DETAILS: Erase all the old equations and sketches. The dusty dark-green chalkboard on the left of frame now carries only this text, in white chalk, in neat but natural hand-written capitals and lowercase, spelled exactly, one line per item, left-aligned, filling the board from top to bottom: Title, larger and underlined: "Using AI to enhance your content" "Step 1: Record clip by clip" "Step 2: Send to Claude to analyze what your clip is about" "Step 3: Have Claude build a still from your clip" "Step 4: Enhance with AI image prompt" "Step 5: Send to editing model to replace what you need" Long steps wrap onto a second shorter line beneath, slightly indented. The chalk has real texture: slightly uneven strokes, faint smudges and ghosts of old erased writing behind. A small chalk arrow or tick beside each step number is fine. No other words anywhere on the board. The chalk tray with chalk stubs and the felt eraser stays. THE SETTING: Same room as the references: the tall polished wooden double doors, the carved white plaster frieze above, the cream tile floor and the red mat at bottom left, with the lectern and chalkboard exactly where they are in the second reference. THE LIGHT: KEY: the same warm overhead light as the second reference plus the green-shaded lamp glowing on the lectern. FIELD: same brightness as the reference. The chalk text reads bright white with a faint warm cast from the room. Colour grade unchanged. THE COMPOSITION: Same framing as the second reference; the full chalkboard text is in frame and legible at phone size, nothing cropped. THE CAMERA: Shot on an iPhone, same lens, same distance and same slight angle as the references, natural phone sharpness and noise.
Exact prompt
Create a new video where the man in @video_1 talks to camera with his exact motion and timing, but he is now the university professor from @image_1, mid-lecture: brown herringbone tweed blazer with suede elbow patches, burgundy sweater vest, pale blue oxford shirt and green striped bow tie, standing behind the carved oak lectern from @image_1 with its brass gooseneck mic, green-shaded lamp, leather books, open notebook and mug, and the green chalkboard with its white chalk lesson standing on the left of frame. His motion, timing, framing and camera stay exactly as in @video_1. [SCENE 1] The outfit, lectern and chalkboard from @image_1 are in place from the first frame and stay put for the whole clip. He keeps every small move of @video_1 exactly: his raised hand at his chin holding the small dark microphone between his fingertips exactly as in @video_1, the small head turns and nods as he talks, his other hand resting on the lectern edge. The tweed blazer moves like heavy wool, creasing at the raised elbow so the suede patch shifts with the arm. The lectern, the books, the mug and the chalkboard are solid and completely still; the chalk writing on the board stays exactly as in @image_1, same words, same letters, frozen in place, never changing, flickering or rewriting. The green lamp glows steadily, casting a warm pool on the books and notebook. The warm overhead room light stays constant. His face stays preserved exactly as in @video_1, no aging, no glasses. Lips, jaw and speech timing exactly as in @video_1, do not alter the mouth shapes. The room stays unchanged: the tall wooden double doors, the carved white plaster frieze above, the cream tile floor and the red mat at bottom left. Shot on an iPhone, same lens, same distance, same handheld camera as @video_1, original audio kept.
6. Line them up
Once all six are done, I put each original clip next to its AI version in Edits. I add a label on each side and a hook at the top, then cut them together in the order of the script.
Why it works
Every costume starts from a real frame of my own clip, and the video model keeps my real motion underneath it. The room, the light and my voice stay the same from clip to clip, and the outfit is the only thing that changes.
Your turn: write three short lines you'd say to camera, and film each one in the same spot with your hands doing something different.
Emero watches your clips, builds the stills and runs the edits for you, and it remembers your looks for the next reel, all at cost.
Get one of these every day
I test new workflows daily and send them with the exact prompts, the models and the costs
6,546 generations tested so far
More workflows
See allKling 4.0 + GPT Image 2.6 Flare - fire, water, earth and air VFX from your own clips
Add fire, water, earth and air VFX to your own phone clips while keeping your real motion
GPT Image 2.6 Flare, Kling 4.0 · 8 min render
Kling 4.0 - AI video effects from your own clips: a stretchy arm, a kiss, a fly and a whip
Four AI video effects from my own phone clips: a stretchy arm, a kiss, a fly, a whip-arm
GPT Image 2.6 Flare, Kling 4.0, Nano Banana Pro · 7 min render
Kling 4.0 edit mode: video to video AI that swaps your head mid-sentence
Feed in a clip you filmed and get it back with your head swapped, voice and lip sync kept
GPT Image 2.6 Flare, Nano Banana Pro, Kling 4.0 · 4 min render