
To do the Migos AI trend, prepare one clear photo per performer and use a two-person AI video workflow. Choose a preset reference performance or generate a new orange-stage duet, check the settings and cost, then review and download the result. In Duet Booth, upload two separate photos, choose Original moves or New performance, and find the finished clip in History.
The recognizable setup comes from Quavo and Takeoff’s “HOTEL LOBBY” performance for A COLORS SHOW: two performers, an orange set and one hanging microphone.1 Motown and COLORS launched their performance series with Quavo and Takeoff on June 17, 2022.2
“Migos AI” is used for AI videos that place a different duo into an inspired performance. Dreamina’s own Migos AI guide uses this name for the orange-booth format.3 For more background, read our Hotel Lobby AI trend guide.
A dedicated generator provides a preset workflow; a general image-to-video tool lets you describe the scene yourself. Dreamina publishes a two-photo, prompt-led approach with Seedance 2.5.3 Choose based on whether you want to manage prompts and scene direction or begin with an existing reference setup.
Duet Booth offers both Original moves and New performance in one workspace. The steps below describe that product’s current controls, checked on October 4, 2026.4 Neither mode guarantees exact appearance, movement or sound.
Go to Migos AI and use the Create tab. The workspace starts with Original moves and the Hotel Lobby reference workflow. You can inspect the settings before creating an account or buying credits.4
Decide which person or pet should appear on each side before uploading.
Select Left performer, then Right performer, and add a separate photo to each slot. Use JPEG, PNG or WebP files up to 10 MB each.4 Both subjects should be easy to identify: clear faces, reasonable lighting and one main subject per image.

Actual Duet Booth workspace showing separate photo uploads. Check the current availability and credit quote in your workspace before creating.
Review both previews before continuing. If the order is wrong, replace the relevant upload. Avoid using the same group photo twice. Remove screenshot overlays and crop out unrelated subjects while keeping useful face and clothing detail. Only upload photos you have permission to animate.
Keep Original moves if you want the preset Hotel Lobby reference as the starting point. You do not need to upload a reference clip or enter theme text. The workspace handles that configured workflow.
Choose New performance for a fresh duet. Select a stage, choose a model and optionally describe an occasion in the theme field. A short theme such as “a playful birthday duet” gives direction; it does not guarantee specific names or word-for-word lyrics.

New performance exposes model selection. Original moves does not display a model selector.
Choose the duration, resolution and format currently supported by your selected mode and model. The workspace offers 5, 10 or 15 seconds and either landscape 16:9 or portrait 9:16. These are Duet Booth options, not specifications shared by every tool.4
Read the credit quote after changing settings. Press Create video when both photos are ready. If you are signed out, complete sign-in; if you need credits, review the purchase options. Buying credits does not automatically generate your video. Return to the workspace, confirm the current quote and submit the task yourself.
Open History under the account that created the task to check its progress. You can leave the page and return later. Check the existing task before submitting another.
Once complete, preview the whole clip and use the available download control to save it. Check the downloaded file before sharing. For failed tasks or download problems, see the detailed product walkthrough.
| Your goal | Choose | What you control |
|---|---|---|
| Use the preset Hotel Lobby reference workflow | Original moves | Available duration, resolution and format |
| Create a fresh duet on a chosen stage | New performance | Model, stage, optional theme and supported video settings |
New performance currently offers Seedance 2.0 Mini, Fast, Seedance 2.0 and Seedance 2.5. Mini and Fast support 480p and 720p; the other two also support 1080p. Use the current options and quote in the workspace. Original moves does not show a model selector.4
Watch the entire clip with sound. Check for swapped faces, merged subjects, abrupt motion and unsuitable audio. Details may change after the opening frame.
An AI-generated video does not give you rights to someone else’s likeness or a commercial recording. If you add music, use audio you are allowed to publish. Make the AI nature clear where viewers might mistake the clip for a real performance.
No. Use two separate images, one per performer. People and pets can be paired without taking a new photo together.
No. Duet Booth’s Original moves mode uses a configured reference. New performance also starts from your two photos without asking you to upload a reference video.
Check the current workspace quote. Cost depends on the mode, model and selected specifications. Compare that amount with your balance before submitting.
No. Payment adds the purchased credits. Return to the workspace, review your photos and settings, then submit the generation yourself.4
Open the Migos AI generator when your two photos are ready.
[1] COLORSxSTUDIOS — Quavo & Takeoff: HOTEL LOBBY, A COLORS SHOW. Official performance; source for the performers, orange set and shared microphone.
[2] Universal Music Canada — Motown & COLORS announce their performance series. Published June 17, 2022; confirms the opening performance by Quavo and Takeoff.
[3] Dreamina — Migos AI Video Generator. Official product guide describing the format and a two-photo, prompt-led Seedance 2.5 approach. Its product claims are separate from Duet Booth’s capabilities.
[4] Duet Booth — Migos AI workspace. Product controls checked October 4, 2026: photo limits, modes, models, specifications, quotes and History. This verifies our interface, not guaranteed generation quality; current availability and pricing take precedence.
