AI video prompts: pour giant coffee, stretch a leg, fly a propeller hat and spin your own tornado, from your clips

Four AI video prompts that add one impossible thing to a plain clip of yourself.

9 generations$2.69 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

This is the video I posted on Instagram, here's the link: https://www.instagram.com/reel/Df5GN-nMBBm/

In this post you'll learn how to:

  • pour litres of coffee from two giant cups in your own hands and keep them pouring the whole clip
  • stretch one leg clear across the screen and snap it back as you kick
  • strap on a propeller hat and actually lift off the ground, floating from the very first frame
  • stand inside a tornado you control and float in the middle of it like a weather superhero

I took four plain clips of myself standing around outside and gave myself four powers I definitely don't have in real life. Giant coffee cups, a rubber leg, a propeller hat that flies, and my own personal tornado. Every one started as a boring 3 to 4 second phone clip of me in a park or a plaza. None of this is a filter you download, it's one AI edit added to my real footage, so the background, my clothes and my movement all stay exactly as I filmed them.

Here's my exact process.

1. Pour giant cups of coffee from your own hands

First I filmed myself standing in a plaza and spreading my arms out wide, nothing in my hands. Then I grabbed the frame where my arms are fully open and built a still that drops a comically huge coffee cup into each hand, both tipping over and pouring.

A man in a plaza holds a giant coffee cup in each outstretched hand, both pouring coffee onto the ground., made with GPT Image 2.6 Flare

Exact prompt

Photoreal edit of the reference photo. Keep the reference photo exactly as it is: same man, his black t-shirt, dark pants, sneakers, his stance with both arms spread wide out to the sides, the same plaza, the restaurant, cars, light and framing. Preserve his face and glasses exactly, no aging. Add ONE thing: a comically giant cup of coffee held in each outstretched hand. THE VIBE: A funny visual hook: he holds an enormous cup of coffee in each hand, arms spread wide, pouring litres of coffee down onto the ground. Photoreal, oversized and absurd. THE SUBJECT: Keep the man exactly as in the reference, arms out wide. In each of his outstretched hands, place a comically oversized takeaway paper coffee cup, each about the size of a bucket, tilted so a thick heavy stream of dark coffee pours out of each one down toward the ground and splashes on the pavement. Two giant cups, one in each hand, both pouring at once. THE FACE: Preserve exactly from the reference, glasses on, no aging. THE SETTING: The same plaza, restaurant, palms, cars and daylight as the reference. THE LIGHT: The same bright daylight; the cups and coffee streams lit to match. THE KEY DETAILS: Two comically huge coffee cups, thick streams of coffee pouring down from each and splashing on the pavement. THE MOOD: Funny, absurd, deadpan. THE COMPOSITION: Keep him centered, arms wide; a giant cup pouring at each hand. THE CAMERA: Same phone capture and framing as the reference. No text, no logos.

The one thing that makes or breaks it: say the cups are locked in your hands from the very first frame and keep pouring the whole time. If you just say "add coffee cups" they pop in and out or vanish halfway. I pin them in place and say the coffee gushes down and splashes on the ground the entire clip. Then I run the clip and the still together in edit mode, so my real arm movement stays and the cups ride with it.

Kling 4.0 $0.61

Exact prompt

[SCENE] Keep everything from @video_1 exactly: the man, his face, glasses and black t-shirt (no aging), the plaza with the restaurant, the daylight, the camera and the timing of his motion. From the very first frame the two comically giant takeaway coffee cups are locked in his hands, exactly like @image_1, one in each hand. As his arms spread wide, a thick heavy stream of dark coffee pours continuously out of each cup straight down, splashing and pooling on the pavement. The cups stay locked in his hands the whole clip and never disappear; the coffee keeps gushing down the entire time. @image_1 is the look to match. Keep his face exactly as in @video_1, no aging.

2. Stretch your leg across the whole screen

For this one I kicked one leg forward and let the camera run. I pulled the frame at the top of the kick and built a still where that leg stretches out like rubber, all the way to the far side of the frame with my shoe at the edge.

A man mid-kick whose leg stretches across the whole frame like rubber, shoe at the far edge., made with GPT Image 2.6 Flare

Exact prompt

Photoreal edit of the reference photo. Keep the reference photo exactly as it is: same man, his black t-shirt, dark pants, sneakers, his profile stance mid-kick, the same palm plaza, the light and framing. Preserve his face and glasses exactly, no aging. Change ONE thing: his kicking leg. THE VIBE: A funny visual hook: his kicking leg stretches and elongates like rubber, reaching way across the whole screen. Photoreal, cartoon-stretchy. THE SUBJECT: Keep the man exactly as in the reference, his upper body, arms and standing leg the same, caught mid-kick. His raised kicking leg, the forward one, stretches and elongates dramatically, about four to five times its normal length, thin and rubbery, reaching all the way across to the far opposite side of the frame, with his sneaker at the far edge. The stretched leg is clearly his own leg, same dark pants and sneaker, just elongated like rubber. THE FACE: Preserve exactly from the reference, glasses on, no aging. THE SETTING: The same palm plaza and buildings as the reference. THE LIGHT: The same bright daylight; the stretched leg lit to match. THE KEY DETAILS: The kicking leg elongated rubber-like all the way across the frame, sneaker at the far side. THE MOOD: Funny, surreal, cartoon-stretchy. THE COMPOSITION: Keep his body on the right; the stretched leg reaches across to the far left edge. THE CAMERA: Same phone capture and framing as the reference. No text, no logos.

In the video I tie the stretch to the actual kick: the leg extends as I kick out and springs back as I lower it. Because edit mode keeps my real timing, the stretch lands right on the motion instead of floating around on its own.

Kling 4.0 $0.60

Exact prompt

[SCENE] Keep everything from @video_1 exactly: the man, his face, glasses and clothes (no aging), the palm plaza, the daylight, the camera and the timing of his motion. As he kicks his leg forward, his kicking leg stretches and elongates like rubber, extending far across the whole frame to the far opposite side exactly like @image_1, his sneaker reaching the far edge. The leg stretches out as the kick extends and springs back as he lowers it, following his real kick motion and timing from @video_1. Keep his standing leg, upper body and face exactly as in @video_1, no aging.

3. Strap on a propeller hat and take off

I filmed myself just standing still in a park. I built a still that adds a colorful kids' propeller beanie on my head and lifts me a couple of feet off the grass, feet dangling, with a shadow underneath to sell it.

A man wearing a colorful kids' propeller beanie floats a couple of feet above the grass in a park., made with GPT Image 2.6 Flare

Exact prompt

Photoreal edit of the reference photo. Keep the reference photo exactly as it is: same man, his navy t-shirt, jeans, sneakers, the same grassy park with trees, the daylight and framing. Preserve his face exactly, no aging. Change two linked things: add a kids' propeller beanie hat on his head, and lift him off the ground so he floats. THE VIBE: A funny visual hook: he wears a colorful kids' propeller beanie (the classic rainbow-panel cap with a little propeller on top), and the propeller is lifting him up so he floats in the air above the grass. THE SUBJECT: Keep the man exactly as in the reference. Put a classic kids' propeller beanie hat on his head: a round multi-colored rainbow-panel cap with a small button on top holding a little two-blade propeller. Lift his whole body about two feet off the grass so he is floating in the air, upright, feet dangling just above the ground, arms relaxed at his sides, as if the hat's propeller is holding him aloft. THE FACE: Preserve exactly from the reference, no aging. THE SETTING: The same park, trees, grass and daylight. A soft shadow on the grass beneath his dangling feet shows he is off the ground. THE LIGHT: The same bright daylight; the hat and his body lit to match. THE KEY DETAILS: A colorful kids' propeller beanie with a little propeller on top, him floating about two feet above the grass, feet dangling, a shadow below. THE MOOD: Funny, light, whimsical. THE COMPOSITION: Keep the framing; he hovers a bit higher than he stood, centered. THE CAMERA: Same phone capture and framing as the reference. No text, no logos.

The tip that matters here: say you're already in the air from the very first frame, feet never touching the ground, and that you stay floating the whole clip. If you don't, the edit keeps you standing and only lifts you off at the very end. Spell out that you're airborne from frame one and the propeller holds you up the entire time.

Kling 4.0 $0.67

Exact prompt

Create a new video where the man is already floating high in the air above a grassy park from the very first frame, held up by the fast-spinning propeller on his colorful kids' beanie hat, hovering and bobbing the whole time. [SCENE 1] Keep from @video_1: the man, his face and clothes (no aging), the grassy park, the trees, the daylight, the camera and framing. From the VERY FIRST FRAME he is already airborne, floating a few feet off the grass exactly like @image_1, feet dangling, and he never stands on or touches the ground at any point in the clip. He wears the colorful rainbow propeller beanie from @image_1; the little propeller on top spins fast the entire time, a blur of rotating blades, lifting him. He hovers and bobs gently up and down in the air, floating steadily, arms relaxed and swaying slightly, his feet dangling the whole clip. His soft shadow sits on the grass below him. Keep his face exactly as in @video_1, no aging.

4. Float inside your own tornado

I filmed myself turning on the spot in a plaza. I built a still that puts me in the middle of a dark tornado, lifted off the ground with my eyes glowing white, papers and debris flying around me, like a weather superhero.

A man levitates inside a dark tornado with glowing white eyes and papers flying around him., made with GPT Image 2.6 Flare

Exact prompt

Photoreal edit of the reference photo. Keep the reference photo exactly as it is: same man, his black t-shirt, dark pants, his stance facing the camera, the same plaza and modern buildings, the framing. Preserve his face and glasses exactly, no aging. Transform the scene so he looks like a weather-controlling superhero (Storm from X-Men) LEVITATING inside a tornado he commands, fully photoreal and integrated, NOT a pasted overlay. THE VIBE: A powerful cinematic hook: he is a storm god floating in the air, in command of a tornado, like Storm from X-Men. He levitates off the ground, his eyes glow bright white with power, the wind whips his shirt and hair, and a dense tornado swirls around him with papers and debris flying. It must look real and fully integrated, like one single photo, never a flat cut-out tornado pasted over a sunny photo. THE SUBJECT: Keep the man exactly as in the reference, facing camera, arms slightly out and powerful, but now LEVITATING a couple of feet off the ground, his whole body floating in the air, feet dangling with toes pointed slightly down, clearly not standing on the ground. His eyes glow solid bright white (Storm's power). His t-shirt and hair are blown hard by the wind, fabric rippling. A dense grey tornado swirls tightly around him from the ground up, wrapping around and partly behind him, with lots of white papers and debris spiraling in the wind. THE FACE: Keep his exact face, bone structure, beard and glasses from the reference, no aging, no change to his features. ONLY his eyes glow white. THE SETTING: The same plaza and buildings, now under a dark dramatic storm, the sky overcast and moody. THE LIGHT: Cool, dramatic overcast storm light falling on EVERYTHING including the man, matching the tornado, so the man and the storm share the exact same lighting and he is clearly lit by the storm and not by bright sun. Rim light from the swirling debris. THE KEY DETAILS: Levitating off the ground, glowing white eyes, wind whipping his clothes and hair, a dense integrated tornado around him, papers and debris spiraling. THE MOOD: Epic, powerful, cinematic, dangerous. THE COMPOSITION: Him centered and floating, the tornado and debris wrapping the whole frame. THE CAMERA: Same phone capture position and framing as the reference, now in storm light. No text, no logos.

Two things make this read right. One, keep your feet dangling and off the ground so it looks like you're floating, not spinning on your feet. Two, say the tornado is violent and the debris is whipping around you, or you get a weak little dust cloud instead of a storm. Edit mode keeps my real turning motion, so I rotate inside the storm on my own spin.

Kling 4.0 $0.61

Exact prompt

Create a new video where the man levitates high off the ground inside a violent tornado he commands like Storm from X-Men, his whole body spinning fast in the air as the storm rages and debris flies everywhere. [SCENE 1] Keep the camera and the timing of his motion from @video_1, and keep his face, glasses and black t-shirt exactly, no aging. He is levitating well off the ground, feet dangling limp and still, exactly like @image_1. His whole body spins fast as one, floating and rotating in the air like a top; his legs hang loose and never step, walk or pivot, and there is no ground contact at any point. Around him a violent, aggressive tornado roars: a dark churning grey-black funnel spinning fast, fierce wind, and a huge amount of papers and debris whipping violently all around him, the ground torn up at the base. His shirt and hair whip hard in the wind, his eyes glow solid white. Dark, dramatic, dangerous storm light on everything, matching @image_1. The spin and the storm grow more violent through the clip. Keep his face exactly as in @video_1, no aging; only his eyes glow white.

Why it works

Every one of these is the same move: take a plain clip of yourself and add exactly one impossible thing on top, in edit mode so the model keeps your real footage. My camera, my timing, my clothes and my background all stay. That's why a giant coffee cup or a tornado sits in the shot like it was really there instead of looking like a sticker pasted on. I build the look as a still first, cheap, get it right, then spend on the video once I know the still is good.

You can do this with any clip of yourself. Film 3 to 5 seconds of one clear action outside, pick the one impossible thing you want to add, build the still, then animate it. Keep each clip short, test the still before you pay for the video, and stitch them together at the end.

Your turn: grab one 3 second clip of yourself doing one simple move and add one impossible thing to it. Start with the coffee, it's the most forgiving.

Emero runs all of this at cost and remembers your face between clips, so the next hook is faster to make.

Here's the full cut one more time:

Total for this workflow$2.69

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