Gangnam Style AI Prompt (Copy and Paste) for the Horse Dance

October 25, 2026

Gangnam Style AI horse dance: a person with crossed wrists on imaginary reels — A prompt to describe the horse dance.

A good Gangnam Style AI prompt names the camera, the place, the dance beats and the face rules. Paste the prompt below into an AI video tool that accepts a reference photo, check the result against the fix list, and add PSY’s 2012 song when you post. The Yuppop generator needs no prompt at all.

Key takeaways

Describe camera, place, beats and face rules.
Always say empty-handed, or the tool draws a rope.
Always say keep the face and clothes from the photo.
The generator runs all this for you.

What is a Gangnam Style AI prompt?

It is a written instruction that tells an AI video tool to make the horse dance from your photo. It fixes five things: the camera (a still phone, head to toe), the place (a golden-hour rooftop), the dance (gallop, then lasso, on the beat), your identity (same face, hair and clothes in every frame) and the failure list. Two or three runs is normal. The AI wave began when @mxna.mxna posted an AI remake on 12 October 2026, days after @emiliyyam’s Seoul reel passed 17.5 million views.

The copy-and-paste prompt

Attach one clear photo of the person as the reference image, then paste:

A vertical phone video, photoreal live action in full natural colour, one continuous take from a phone on a tripod at chest height. The dancer is seen head to toe the whole time, filling most of the frame height. The camera does not move. No cuts. No text, no subtitles, no logos.
Setting: a city rooftop terrace at golden hour: warm low sun from one side, a soft orange and pink sky, glass towers in the distance, a low concrete parapet and a few potted plants behind the dancer, a string of warm bulbs overhead. Nobody else is there.
Who: one dancer, the person from the reference photo, in exactly the clothes they wear in that photo, alone, empty-handed, facing the camera the whole time.
Action: the Gangnam Style horse dance, on the beat, dancing in place, in this order. First, they stand in the middle of the frame, knees bent, wrists crossed low in front of the belly, hands empty, with a deadpan stare and a confident grin. Next, the gallop: wrists stay crossed, they bounce from foot to foot on every beat, knees lifting. Then the lasso: one empty hand circles above the head with nothing in it, the other wrist stays at the belly, and the feet keep bouncing to the last frame.
Identity: the dancer is exactly the person from the reference photo in every frame: the same face, age, skin tone, hairstyle, hair colour and hair length, the same clothes and the exact body build, never slimmer, heavier or younger.
Avoid: walking off, turning away from the camera, another person, a horse, a rope, reins or anything held in the hands, a moving camera, close-ups, face changes, the dancer talking, captions, text, logos, cuts, extra arms, extra fingers, distorted hands, morphing faces.

If your tool does not name reference images, add “the person in my reference photo” where the prompt says “the person from the reference photo”, and turn the tool’s own music off. Want it without prompting? The Gangnam Style generator makes it from one photo.

What does each part do?

Camera and setting. One continuous take, head to toe, camera still. A tool that zooms or cuts away loses the feet, and the feet are the gallop. The rooftop is described piece by piece so the tool builds it instead of guessing.
Who and avoid. You, in the clothes from your photo, alone and empty-handed. “Empty-handed” exists for the lasso: without it tools hand you a real rope.
Action. The steps are in order, with no time stamps, so a tool with any clip length can follow them. “Dancing in place” stops the tool walking you around.
Identity. Bouncing is where faces, hair and builds drift, so the line repeats itself on purpose.

How do you fix a failed run?

A real rope in the lasso hand: keep “one empty hand, nothing in it” and put a rope under things to avoid. Never write “as if swinging a rope”.
The camera zooms or moves: keep “The camera does not move” and “a moving camera, close-ups” in the avoid list.
The face drifts: repeat the identity line and use a sharper, front-facing photo.
Extra limbs or fingers: say exactly two arms and two legs and five fingers on each hand, then retry.
Text appears: repeat “no text, no logos”.

Can you adapt it for a pet?

Pet clips are part of the trend. Replace the dancer with one animal from your reference photo, standing upright, and keep the same order of beats. Remove the human-only lines about clothes and hair, and keep the identity and avoid lines.

What is the easy route?

Skip prompting: use the generator with one photo chosen with the photo guide. To see the moves the prompt describes, read the step-by-step guide.

Yuppop is an independent tool and is not affiliated with PSY, his label or any social platform. Make your own take with the Gangnam Style AI video generator.

FAQ

What is the prompt for the Gangnam Style trend?

Describe a vertical phone video on a sunset rooftop where the person from your reference photo does the horse dance: wrists crossed like reins, a gallop in place, then a lasso swung with one empty hand. The full copy-ready prompt is in this guide.

Do I need a prompt on Yuppop?

No. The generator sends its own, so you only upload a photo.

Why is there a rope in my video?

Tools read “lasso” and draw one. Keep “empty-handed” in the prompt, say “nothing in the hand” and list a rope and anything held in the hands under things to avoid.

Will the prompt add the song?

No. There is no sound line, so you add PSY’s song in the app when you post.

What if the hands or face look wrong?

Repeat the face rule, ask for five fingers on each hand and run again. Every run is a new take.