Gangnam Style AI Prompts: Roles, Motion and Reference
Our generator already supplies its prompt and dance reference. These examples explain the instructions behind a three-person video and can help when working with a separate model that accepts image and video references.
Give each person a place in the trio
Photo 1 goes on the left, photo 2 in the center and photo 3 on the right. The center performer gets the close camera moments, while the people on either side have their own turns. Check the slot labels before generating. Each photo guides that person’s appearance; the reference video guides the performance.
Describe motion separately from appearance
Use the portraits for identity and the reference clip for timing, staging and camera movement. Asking only for “Gangnam Style dancing” leaves the model to invent the scene. Name the reference input explicitly, and ask it to preserve the sequence of close-ups and individual turns.
Set limits and review the result
Ask for the same three people throughout, with stable roles and clothing, without extra performers or added text. A prompt expresses your intent; it cannot guarantee exact faces or perfect hands. This site uses one fixed scene and does not accept custom prompts. Copying this example alone does not supply a reference video to another tool.
Recast the three performers in Video 1 using the three portrait references. Image 1 is the left performer, Image 2 the center and Image 3 the right. Keep each identity, hairstyle and outfit consistent across shots. Follow the reference camera turns, close-ups and gestures at their original timing. Keep the reference setting. Use only these three performers and omit added text. Output a 15-second landscape clip in 16:9.