The manual way: build it in any AI tool
If you prefer to work in your own image and video tools, split the job in two. Most failed attempts happen because people ask a video model to invent the room and keep two faces at the same time. Building the still picture first gives you control over who stands where and what the booth looks like.
Step one is the scene. Put the photos side by side into an image editing model and describe the room, the framing and what must not change. Step two is the motion. Give that picture to an image to video model, describe what each performer does, keep the camera still and ask for vertical 9:16 with sound if the model supports it.
- Scene prompt we use for Duet: "Take the two people from the left and right halves of this image and show them standing side by side together in a minimal studio booth with smooth glowing orange walls and a single microphone hanging from the ceiling between them, warm orange light. Keep both faces, hairstyles and outfits exactly the same as in the photos. Photorealistic, both people visible from the knees up, vertical 9:16 frame, no text."
- Motion prompt we use for Duet: "The two people trade rap lines into the hanging microphone between them, one leans in to rap while the other nods and moves to the beat, then they switch. Warm orange light, static camera, music performance video look. Keep both faces, hairstyles and outfits exactly as in the photo. Natural, realistic motion."
- Keep the camera static. Camera moves are the fastest way to lose the faces.
- Ask for an original instrumental beat with no vocals. Real songs and artist voices are a copyright problem, and you can add a trending sound in the app when you post.
What you get, free and paid
Free clips are 480p and 4 seconds: enough for one moment at the mic. You get 1 free video with no account, and a free account adds 1 more. It is a one-time allowance, not a daily one.
After that, a Hotel Lobby clip costs 50 credits, because it takes two AI steps (the scene picture and the video with sound). An HD clip at 720p and 8 seconds costs 80 credits and leaves room for a real back and forth between two performers. Credit packs are one-time payments, with no subscription, and credits last 12 months.
Tips for posting
The trend works because of the pairing, not the rapping. Say who is in the booth in the first second of the caption: "me and my grandma", "my cat finally dropped her album", "the two interns". Post the clip as it is so the booth fills the screen, and keep the caption short so it does not cover the faces.
The clip comes with its own beat. On TikTok and Instagram you can also pick a trending sound in the app; if you do, lower the original sound so the two do not clash.