Make your Gangnam Style AI Video in four steps
Choose your trio
Start with a clear adult portrait for the camera holder. Add one or two friends if you want to choose the whole cast; fictional adult companions fill the empty positions. Separate photos work well, so everyone can send a favourite selfie from their own phone. Choose portraits you have permission to use.
Set the photo order
Match each thumbnail to its role. The first person holds the camera, the second begins on the viewer's right and the third on the viewer's left. Use the move buttons to swap places or replace a photo. Removing a portrait moves the following photos forward. On desktop, the phone-upload option lets you bring pictures over from your phone.
Create the video
Sign in and select Create after checking your cast. The template supplies the performance, setting and music, so your photos are all you need to prepare. You can leave the page while the video is being made and return with the same account to check progress.
Watch with sound and save
Enable sound in the player and watch each person's turn through to the group ending. Then download the finished MP4. Keep the landscape version for a group chat, or make a separate crop for a vertical feed. Before publishing, check the destination platform's rules for using the song.
The trio selfie behind the Gangnam Style AI trend
The Gangnam Style AI video format puts three people in a handheld selfie set to PSY's song Gangnam Style. A shared opening leads into individual close-ups, with a quick camera move handing the spotlight to the next friend. The trio comes back together for the finish. The fun comes from the changing expressions and the sudden arrival of a familiar face, rather than a full-body horse-riding dance routine.
Our real-life reference is @emiliyyam's TikTok post captioned “gangnam style in gangnam”. Three friends perform on a street at night, close enough for the camera to move between their faces. Watch the linked post to follow the complete sequence, then use the fifteen-second reference above to decide who you would place in each position. Notice the camera holder's role: that person brings the group together and remains the visual anchor as attention shifts to the others.
The night-street AI example by @bikonyavamm recasts the trio with the likenesses of Turkish actors and retains the creator's visible watermark. The snowy example by @yapayzekaserisi is a Harry Potter-inspired AI adaptation with Harry, Ron and Hermione. On this page, your uploaded portraits become the cast of the fixed night-street selfie; the examples are viewing references rather than scene choices.
For a friends' version, think about the reveal you want. Put the person who usually takes the group selfie in the first position, then give the other two their own turns. Or choose an unexpected camera holder and save the most expressive friend for the final solo moment. Those choices are made through the photo order, so you can plan the cast before pressing Create. The visible labels are especially useful when the photos came from different devices and have similar filenames.
Watch the reference once for the whole group, then once more for the camera movement. Look at when a face fills the frame, how long the next person has to react and how the group returns for the ending. That gives you a practical way to review your own clip later: the people should be easy to follow, the solo turns should feel distinct and the final shot should bring the trio back into one composition. The rhythm of the handoff is what makes this particular selfie format work.
Watch the real-life trio selfie on TikTok ↗AI adaptation by @bikonyavamm using Turkish actors' likenesses for a night-street trio; the creator's watermark is retained.
@bikonyavamm · TikTok · View TikTok source ↗Choose portraits that work in a moving close-up
Choose photos with enough facial detail to stay clear in a close-up. A front-facing selfie or a gently turned head is a useful starting point. Look for even light across the eyes, cheeks and mouth, with the hairline visible. A window-lit portrait often makes it easier to see those details than a dim party picture. If you are choosing between several photos, enlarge each one and compare the face rather than judging only the thumbnail.
Give every person their own picture. A clean portrait helps you keep track of the three roles and makes replacements straightforward. Check that the mouth is visible, the face is in focus and accessories leave the main features readable. Similar brightness across the portraits can make your selection easier to assess together. You can mix vertical and horizontal source pictures; the template creates a landscape video from either orientation.
Clothing and hairstyle are useful visual cues when the camera moves quickly. Pick a photo that looks like the person your friends expect to recognise: their usual hairstyle, a familiar jacket or a recent selfie. If one picture has a very strong beauty filter, look for a more natural alternative with clearer features. The aim is a trio that feels like the same three people from the opening through the last group shot.
Use the cast controls as a final rehearsal. Read the camera-holder label first, then check the right and left positions. After moving or removing a photo, look at the labels again because the remaining pictures shift to fill the available slots. If a friend sends a better image while you are choosing, replace that portrait and check the order before creating. This keeps a small photo correction from accidentally changing who appears in the central role.
When your video is ready, play it once for the overall joke and again to inspect each person's close-up. Check the eyes, mouth, hair and hands as the camera changes direction. Enable the player's sound to hear the selected music alongside those turns. If you plan a vertical edit, check every turn in the cropped frame: the person at the side may need room just as much as the camera holder. Save the landscape original so you can return to it for a different crop or a group-chat version.
Harry Potter-inspired AI adaptation by @yapayzekaserisi: Harry, Ron and Hermione take turns in a snowy Hogwarts selfie.
@yapayzekaserisi · TikTok · View TikTok source ↗