Hotel Lobby AI With Your Own Reference Video: How It Works
Yes, your own video can drive the moves. Open Soundtrack & moves, choose Your clip and upload it: the two people from your photos copy its moves, gestures and camera motion on the orange booth or another set, and the clip’s background is swapped out. The video runs as long as the part you use, 2 to 15 seconds; 10 seconds on Wan 3.0 at 480p is 5 credits.
It suits anyone who wants to film their own version of the routine. On October 1, 2026, Victoria Monét and Quavo re-shot the booth themselves with the caption “No AI” (Revolt via Yahoo); a take like that, filmed by you and your friend, can become the movement for your two photos.
Last updated
Which input controls what
| Input | Controls | Does not control |
|---|---|---|
| Your two photos | Who performs: faces, hair, clothes | How anyone moves |
| Your clip | Moves, gestures, head turns, timing, camera motion | Who appears or where: its people and backdrop are replaced |
| The stage | The set: orange booth or another stage | The moves |
| The sound | What the left person lip-syncs to on Wan 3.0, or an AI rap on Seedance | The moves |
Your choreography
Add your reference clip here
Drop in the two photos, then open Soundtrack & moves and pick Your clip. Each length button shows its credits.
Your duo, one photo each
A single person per photo. Left raps, right hypes.
Stage
Video and audio
Cost: 6 credits. Takes roughly 5 to 15 minutes. If it fails, the credits come back. See prices
Filming a clip that copies well
- Two people side by side, full body, facing the lens. The right-hand performer follows the second person in your clip.
- Hold the camera still. Pans, zooms and cuts are copied too.
- Record at normal speed; slow-motion clips do not work.
- MP4 or MOV straight from the phone camera is fine. Clips that are too big or too fast are converted in your browser.
- Any length is accepted; 2 to 15 seconds of it are used.
- Use footage you shot or have rights to, and skip clips of celebrities: the model is told to replace the people in the clip, but faces from a reference can still leak into the result.
Uploading the clip and choosing a sound
- Add both photos, full-length if you can: with Your clip, photos are not redrawn first.
- Press Change on the sound row, choose Your clip under Moves and upload the video.
- If it is longer than the video, slide the orange frame to the part you want. The length buttons show credits for every length your clip allows.
- Pick the sound: Your clip’s sound (chosen first if the clip has audio), the Hotel Lobby track or Your song. A silent clip falls back to the Hotel Lobby track. On Seedance, the sound is an AI rap.
- Check the shape. Choosing Your clip switches it to Match clip, the ratio closest to your footage.
- Press Create video.
What it costs and how long it takes
| Seconds of clip used | Wan 3.0 480p | Wan 3.0 720p |
|---|---|---|
| Up to 5 | 3 credits | 4 credits |
| Around 8 | 4 credits | 6 credits |
| Around 10 | 5 credits | 8 credits |
| Around 12 | 6 credits | 10 credits |
| Around 15 | 7 credits | 12 credits |
A part that falls between two lengths is billed at the next length up. Neither the set nor the sound changes the price. A Wan 3.0 render normally takes about 5 to 15 minutes.
What a reference video can and cannot do
- The prompt asks the model to copy the moves, gestures, timing and camera. How closely it follows varies with the clip, so a simple, steady take is the safest bet.
- Hard cuts in your clip tend to come through as cuts in the result.
- Faces from the clip can bleed into the performers. Use footage of people who agreed to it.
- With only one person in the clip, the right-hand performer has nobody to follow; film two people.
- Your clip’s background never appears: the set replaces it.
Built-in routine or your own clip?
| Stick with the built-in routine if | Upload your own clip if |
|---|---|
| You want the trend’s exact look and choreography | You want a different dance, a skit or your own spin |
| You only have face photos (the redraw adds the body) | You have full-length photos of both people |
| You want the Hotel Lobby track lined up with the moves | Your clip has its own audio, or you are bringing a song |
FAQ
Can my own phone video set the moves for two people from photos?
Yes. Choose Your clip under Soundtrack & moves and upload 2 to 15 seconds of video; the people from your photos copy its moves on the set you pick.
Does my clip’s background show up?
No. The set replaces it; only the movement and camera come from your clip.
Which models accept my own clip?
Wan 3.0 on any set, and the Seedance models when the sound is an AI rap.
Can I keep the audio from my clip?
Yes. Your clip’s sound is chosen first when the clip has audio, and you can switch to the Hotel Lobby track or your own song.
Can I copy a TikTok dance someone else posted?
Only if you have the right to use that footage. Other people’s clips can also carry their faces into your result.
Is a video from my own clip more expensive?
No. It costs the same as any video of that length: 5 credits for 10 seconds on Wan 3.0 at 480p, whatever the set or sound.
References
Up next
Hotel Lobby Trend Tutorial: Pick Your Route, Then Post
A decision-first tutorial for the Hotel Lobby trend: which of three routes suits your time and budget, the steps for each, the free options and how to post
Voice Note to Hotel Lobby AI Video: Your Words, Lip-Synced
When the words must be yours: record on your phone, get the file into the browser, frame the best seconds and let the left performer lip-sync it
Best Photos for a Hotel Lobby AI Video (and Ones to Skip)
Choosing photos for Hotel Lobby AI: the cropped group-chat selfie mistake, what the portrait redraw does, two photos versus one, and what no photo can fix
Hotel Lobby Video is an independent product with no ties to Quavo, Takeoff, Migos, Quality Control Music, Motown or COLORS.