
Video SwapRegenerateTwo cats take the orange booth
Viral trend · Quavo & Takeoff, COLORS 2022
The reference clip is already loaded and the swap instructions are built in. Add two photos, one for each performer, and generate. People and pets both work.
Review results, manage assets and access generation history in AI Studio.
Generated on ImgAI from two cat photos and the muted reference clip. Tap the example to load it into the composer.
Every clip below is a real job run here. Tap one and the prompt and stills load — run it yourself.

Video SwapRegenerateTwo cats take the orange booth
Three things to know before you make one.
A flat orange backdrop, one microphone hanging from the ceiling, two performers trading lines straight to camera. That framing is the whole meme.
Quavo and Takeoff performed Hotel Lobby on A COLORS SHOW in June 2022. In September 2026, TikTok clips started putting cats, celebrities and friends into that booth, and the format spread.
A ten-second muted excerpt is the motion and scene reference. The model keeps the booth, the mic, the timing and the outfits, and swaps in your two subjects.
The reference clip sits in the video slot and the swap instructions are built into the scene. The prompt box stays empty unless you want to add a request.
The first photo replaces the performer on the left, the second the one on the right. One subject per photo, face visible.
The clip follows the reference, about 10 seconds, and lands in AI Studio. Download it and add the track in your editing app; the original audio is not part of the render.
A front or three-quarter view with the face unobstructed. Hats and glasses in the photo get replaced by the performer's sunglasses anyway.
Daylight or a plain indoor light. Heavy shadows and colour casts show up in the result as a different skin or fur tone.
A group photo makes the model guess who you meant. Crop to one person or one pet per photo before uploading.
The example on this page uses two cats. The model puts the outfit and sunglasses on the animal and keeps its fur and eye colour.
Video generation requires an account and credits. The cost depends on the model, output duration, resolution and inputs such as reference video. Try a shorter clip to check the shot before choosing a longer or higher-resolution version. The generator shows the cost before submission; credits are available through subscriptions or one-time packs.
About 1728 credits at 480p. Signing up gives 50 credits; credit packs are on the pricing page. The exact price shows in the composer before you submit, and a render that fails moderation is refunded.
The reference must be muted: a clip with the original song is rejected by the video provider's audio-copyright check. Add the track afterwards in TikTok, CapCut or any editor.
Only with their consent. Photos of real people go through moderation before generation, and public figures or minors may be rejected. You are responsible for what you upload.
The length follows the reference clip, about 10 seconds, because this is a video edit rather than a fresh generation. Resolution can be raised to 720p or 1080p in the composer; the price changes with it.
Yes. Replace the video in the composer with any clip up to 30 seconds, without a copyrighted soundtrack, and keep the prompt's structure: say who replaces whom.
Yes. Leave the prompt box empty to use the built-in instructions, or write one line, such as a different background colour. It is appended after the built-in instructions and wins where they conflict.