Hotel Lobby AI Explained: Where It Came From and Three Ways to Make a Video

In mid-to-late September 2026, videos of two performers against an orange backdrop started spreading across social media. Some featured fictional characters; others featured animals. The cast changed, but the background, music, and movements looked familiar.

One version featuring characters from Breaking Bad had around 324.5K likes when the screenshot was taken. Another, starring two cats, had roughly 7.8 million views and 73K likes on X.

An AI recreation featuring Breaking Bad characters. Source: Glorify on TikTok.

Another take on the same performance. Source: Amy on TikTok.

The two-cat version shared by @ogfeitoza on X. Social metrics reflect the figures shown in the screenshots.

The interest also showed up on Google. In the worldwide, past-year Web Search view on Google Trends, interest in “hotel lobby ai” rose sharply in late September, reached its peak for the selected period, and then declined.

Source: Google Trends, worldwide, past year, Web Search. The chart shows relative search interest, not the number of searches.

So what are these videos, why did they catch on, and how do you make one?

What Is Hotel Lobby AI?

In 2022, Quavo and Takeoff performed “HOTEL LOBBY” on A COLORS SHOW. The two stood against an orange backdrop, sharing a suspended microphone and taking turns rapping. That performance is the source of the setting and movements you see in these AI recreations. You can watch the original performance here.

A still from the original performance. Source: Quavo and Takeoff, A COLORS SHOW.

A quick clarification: “Hotel Lobby” is the name of the song. The orange space in the video is the performance set.

Hotel Lobby AI refers to using AI to recreate that performance with a different pair of characters, while following the original scene, movements, and rhythm.

Put simply, you choose who gets to appear in the familiar two-person rap performance. It could be fictional characters, you and a friend, or even the two cats shown above.

Why Did It Suddenly Become Popular?

Why would a performance from 2022 become something people want to remake years later? Looking at what these videos have in common, I think a few things help explain it.

First, the scene is easy to recognize. The orange backdrop, hanging microphone, two performers, and familiar music make a distinctive combination. After watching a few versions, you can recognize the format almost immediately. Creators can reuse the setting without having to explain the joke each time.

Then there is the cast. Changing the two performers gives the same scene a different kind of humor. Two cats taking their performance completely seriously are funny on their own. A familiar pair of fictional characters adds another layer because you already know their relationship.

It also makes you imagine your own version. After watching someone else’s clip, you might wonder what it would look like with you and a friend. Especially if neither of you usually sings or feels comfortable in front of a camera, it could be a fun surprise to share in your group chat.

Finally, AI tools make the idea easier to try. Some tools already have the reference performance and generation settings prepared, so you can start with photos. You do not need to film yourself rapping or learn video editing from scratch. The jump from watching a clip to making your own becomes much smaller.

If you already have a pair in mind, here are three ways to make the video.

How to Make a Hotel Lobby AI Video

You can prepare a reference clip yourself and use a general-purpose AI video tool, or use a dedicated tool with the performance already set up.

Here are three methods, with the steps and tradeoffs for each.

Method 1: Adapt the Video with Higgsfield Genjutsu

This method uses your own reference clip and photos. It suits people who want to choose a specific part of the performance or adjust the scene.

1. Download and trim the original video. The full performance is 2 minutes and 46 seconds long. Use CapCut or another editor to extract a 10–15 second section as your motion reference.
2. Prepare two photos. Use one clear, unobstructed photo per person, and decide who goes on the left and right.
3. Open Genjutsu. Choose Motion Transfer to guide the generation with the reference clip’s movements and timing.
4. Upload the materials and assign the roles. For example: “Use the first photo for the left performer and the second for the right performer. Follow the reference video’s movements and rhythm, keeping the orange backdrop and hanging microphone.” Adjust the reference labels to match the interface.
5. Check the settings and generate. Review the credit cost before starting. Then check the people, movement, and sound in the result, and adjust the music if needed.

Pros: You can choose your own clip and adjust the characters or scene, giving you more room to customize the result.

Cons: You need to download and trim the reference video, write instructions, and check whether the soundtrack needs extra work.

Method 2: Use Seedance 2.5 with Photos and a Reference Video

For an English-language starting point, open Seedance 2.5 in Dreamina. Its reference workflow supports images and video, so you can provide both the people and the performance you want to follow.

1. Prepare the reference materials. Trim the original performance to the section you want and prepare two photos. Clear full-body photos can help if you want to preserve each person’s outfit and overall appearance.
2. Choose Seedance 2.5. Open the AI video workspace and use the reference-to-video workflow. Add your trimmed clip and both photos as references.
3. Write the prompt. Explain that the video supplies the movement and timing, and assign each photo to a performer. You can start with the prompt below, adjusting the reference labels to match your uploads.

Use the reference video for the performers’ movements and timing. Place the person from Photo 1 on the left and the person from Photo 2 on the right. Preserve each person’s facial features and clothing. Set the performance in an orange studio with a microphone hanging between them. Use a fixed camera and create one continuous shot of the two people performing together.

4. Generate and check the soundtrack. Review the duration, resolution, and cost before generating. Check the people and movement in the result. If the video does not include the music you want, add the track in CapCut and align it with the performance.

Pros: The photos and prompt let you describe the outfits, scene, and other details you want. If you already use a platform with these reference options, you can stay with a familiar workflow.

Cons: You still need to prepare a reference clip, write a prompt, and check the music. Upload options, settings, and pricing differ between platforms.

Method 3: Upload Two Photos to HotelLobby.studio

If you simply want to make the kind of two-person rap video shown above, you can use HotelLobby.studio. The reference performance and music workflow are already prepared, so there is no need to download and trim the original video.

HotelLobby.studio: upload the left and right performers, choose your settings, and generate.

1. Upload two photos. Put one in Left person and the other in Right person. Use one person per photo. JPG, PNG, and WebP are supported.
2. Choose your settings. Select the model, video length, and resolution. For a first attempt, you can keep the default settings. The generation button shows the credit cost.
3. Start generating. Check the photos, left and right roles, and credit cost, then click Generate Video. Wait for the video generation and music processing to finish.
4. Preview and download. Check the result, download the video with music, and share it on social media or send it to a friend.

Pros: No reference clip to find or trim, no prompt to write, and no separate music-editing step. There is less to prepare and manage yourself.

Cons: The tool is built around the preset Hotel Lobby performance. If you want a different performance clip or more control over the scene and clothing, the first two methods give you more flexibility.

Which Method Should You Choose?

	Genjutsu	Seedance 2.5 reference workflow	HotelLobby.studio
Photos	Bring your own	Bring your own	Bring your own
Reference video	Download and trim it yourself	Download and trim it yourself	Already prepared
Prompt	Describe the changes you want	Describe the people, movement, and scene	No prompt needed
Music	Check and adjust if needed	Check and adjust if needed	Handled within the tool
Best suited to	Choosing your own clip and customizing it	Working in a familiar AI video platform	Making the preset two-person performance

If you enjoy adjusting the scene or want to try a different part of the performance, either of the first two methods gives you room to experiment. If you just want to see yourself and a friend in the video, HotelLobby.studio takes care of the reference clip, prompt, and music workflow. You can start with two photos.