The Hotel Lobby AI Trend, Explained: Where It Came From and How to Make One Without the Song

If your feed has been full of cats, couples and TV characters rapping side by side in front of an orange wall since September, you have seen the Hotel Lobby AI trend. This post covers where it came from, the two very different kinds of video that go by the same name, and how to make one that doesn't borrow anyone's song.

We make a Hotel Lobby AI video generator, so we have a stake in this. Where the facts come from other sources, we link them.

Where it came from

"Hotel Lobby (Unc & Phew)" is a 2022 song by Quavo and Takeoff of Migos, released on May 20, 2022 under their duo name Unc & Phew. In June 2022 they performed it on A COLORS SHOW, the YouTube series known for filming artists against a single flat color. Their color was orange.

The name of the trend comes from the song. Nothing in the video takes place in a hotel lobby.

In mid-September 2026, AI remixes of that performance took off on TikTok and X, replacing the two rappers with cats, couples, best friends and fictional characters. One early version used characters from Suits, according to a write-up of the trend. Celebrities joined in too. On September 23, Quavo reposted the unaltered COLORS video on his accounts, and on October 1 he and Victoria Monét posted their own version, without the AI.

Why this format went viral

The format is easy to recognize from a single frame:

• a flat, saturated orange background
• two performers side by side
• a microphone between or in front of them
• a steady, medium-wide shot with no cuts
• a back-and-forth delivery, where one raps while the other vibes

That's what makes it work as a meme. You can put almost anyone in it and people still know what they're looking at. It's the composition that carries the joke, not the specific song.

Two very different things share the name

This is the part most guides gloss over, and it matters for what you post.

Face swaps of the original performance

Many of the viral clips are face swaps. As Complex describes it, apps such as Starrd and Summrs, along with CapCut templates, map two uploaded photos onto the COLORS performance. The movements, the lip sync and the original audio stay the same. Only the faces change.

That's quick and accurate, which is why it spread. It's also where the criticism comes from:

• Takeoff died in November 2022, months after the performance. Some fans felt that a performance he gave months before his death should be treated as a memory of him, not raw material for face-swap memes.
• The audio is still the song. Uploading a clip built on a commercial recording is the kind of thing that gets muted or taken down, and you don't control which.
• Consent and data. Swapping someone else's face in requires their photo, and people have asked what these apps do with the photos they collect.

Quavo himself has not condemned the trend. He reposted the original with a celebratory caption. But the discomfort is real, and it's worth knowing about before you post.

Original clips in the same style

The other kind is generated from scratch: a new video of two people in an orange studio, rapping a new verse over a new beat. It borrows the look, not the performance or the recording.

You give up the exact moves and the familiar song. In return, the words can be about you and your friend, the audio isn't anyone else's track, and you aren't wearing a face over someone else's performance.

How to make an original Hotel Lobby style video

These steps are for our Hotel Lobby AI video generator, but the photo and lyric advice applies to any tool.

1. Pick two photos. One clear, front-facing photo per person. They don't have to be taken together, but similar framing and light help: two head-and-shoulders shots look like one booth, a selfie next to a group photo does not. Skip sunglasses, masks, heavy filters and side profiles.
2. Decide who goes first. The person on the left raps the first half of the lyrics, the person on the right raps the second half.
3. Choose 10 seconds. Five seconds usually only fits one person's lines. Ten gives each person a couple of bars.
4. Write four lines, or let the AI write them. Two for the left, two for the right. Make the second half answer the first. A claim and a comeback works better than two unrelated verses. If you'd like help, the free rap lyrics generator writes a short, clean verse from a topic.
5. Render and check. You get a vertical 9:16 MP4 with vocals, a beat and lip sync, usually in about a minute.

One difference from the original: in our version each person raps at their own vintage microphone on a floor stand rather than sharing one mic that hangs from above. We changed this on purpose. When we let the video model add the mic, it often ended up sitting on top of someone's head, and a mic dangling into the frame from above looked awkward even when it landed in the right place. The orange wall, the side-by-side framing and the static shot are all still there.

Before you post

• Get both people's OK. Both faces are in the video, so both people should agree, and keep it to people you know. Our generator doesn't accept photos of celebrities, public figures or minors.
• Label it if it looks real. TikTok has an AI-generated content label, and YouTube asks creators to disclose realistic synthetic media. A photo-real clip of two friends rapping is the kind of thing those labels are for.
• Don't add the original song. If you overlay the real track on an AI clip, you're back to posting someone else's recording.

More ideas in the same spirit

If the orange booth is not quite what you want, the same two photos can go into an AI rap battle in front of a crowd, or into a birthday rap for the friend in the other photo. And if you want the full step-by-step for any photo, see How to Make an AI Rap Video From a Photo.