The short answer
The 58th Floor trend is a viral format where two photos become four mirror shots of two people trading bars at one microphone. It started as a song by Cordae and Anderson .Paak, became a COLORS performance, and went viral as an AI video template because two people can recreate it with zero shooting equipment, a selfie camera, and thirty seconds of their time.
Where it comes from
The 58th Floor is the opening track of Cordae's album Heavy Is The Crown, released October 23, 2026.
On October 30, 2026, Cordae and Anderson .Paak performed the song on A COLORS SHOW, the YouTube performance series that has built a format around one-take studio recordings. The performance is shot in a teal studio with a hanging microphone, four camera angles, and a call-and-response structure: one artist takes the hook, the other raps the verse, and the two trade shots for 12 seconds.
By November 2, the performance had seeded enough that multiple platforms released templates. The first major one came from Starrd, which let anyone drop their two photos into the same four-shot setup. Five domains registered within 24 hours. By November 3, the phrase "58th Floor AI" was the top search spike across YouTube and TikTok.
Why it is called that
The 58th Floor is a song title, not a building. The name has no connection to real-estate, no skyscraper banner, and no AI-art origin. The trend borrowed its name straight from the track, and the track's name is a metaphor in the song itself.
It is not the same format as the Hotel Lobby trend (the orange booth), and it is a separate COLORS set. The Hotel Lobby is Quavo and Takeoff from 2022; the 58th Floor is Cordae and Anderson .Paak from 2026.
Why it spread
The 58th Floor format broke through because the bar to entry is lower than any other trend in recent memory.
A normal duet trend requires:
- Two people in the same room at the same time
- A camera or a phone
- A location that looks reasonably good
- Time to shoot, frame the shots right, and edit it together
The 58th Floor requires:
- Two photos, each a normal selfie
- No filming, no location scouting, no editing
- Thirty seconds to upload the photos and press Create
- Nothing else
Because the four shots are generated rather than filmed, the two people do not have to perform at the same time, in the same room, or even in the same year. A photo can be years old. One person can surprise the other by making the clip without warning. The photo quality does not need to be a professional headshot—a phone selfie is enough.
The combination of simplicity and the celebrity origin gave it just enough weight to go viral. By the time TikTok and YouTube caught on, the template and the song had already built momentum without any A-list promotion. Cordae and Anderson .Paak did not push it. It spread itself.
The four-shot structure
The 58th Floor AI video takes your two photos and turns them into a 12-second clip with this structure:
- Hook - two-shot, both faces at the mic
- Verse - close-up of the second person rapping
- Wide - full-length shot of both bodies bouncing along the mic
- Face-off - close-up trading between the two faces
The original COLORS performance establishes the angles, the eye line, and the depth. The AI version follows the same template: the only variable is your two portraits. Everything else—the room, the mic, the camera motion, the choreography—is already built into the set.
The output is a vertical 9:16 MP4, 12 seconds, baked with audio. It posts straight to TikTok, Reels, Shorts, or any other vertical platform. No watermark, no additional editing needed.
The sound and rights
The 58th Floor AI set uses audio generated in the style of the original song, not a sample of the recording itself.
If you choose to make the clip with generated bars (by typing a topic), the audio is produced by the model and written around your prompt. No original recording plays. No sample clears. No rights issue.
If you choose to upload your own audio, it must be something you own the rights to. You cannot use the original Cordae and Anderson .Paak recording or any other copyrighted track.
The generated track includes beat and instrumental, so posting a video generated with AI bars works out of the box. No sound from the original artist appears in the file.
What the 58th Floor AI trend is NOT
Not official. The 58th Floor is an independent tool. It is not made by Cordae, Anderson .Paak, COLORS, YouTube, or anyone associated with the song. It borrows the performance format, not permission.
Not a face swap. A face swap takes a face and pastes it onto existing footage, leaving the body and actions from the original video. The 58th Floor generates a new performance from your two photos—your faces move, react, sing, and get their own screen time. It is generative, not a replacement.
Not the CapCut template. CapCut has trending templates, and one of them uses a similar four-shot edit. But CapCut templates are static overlays. The 58th Floor actually generates new video frames based on your input photos. Different technology, different output, different rights situation.
Not claiming your likeness to sell something. The clips you make do not train any model, do not become training data, and do not get used anywhere except your own post. Each clip is private unless you choose to share it.
Day-by-day timeline
| Date | Event | Impact |
|---|---|---|
| Oct 23 | Cordae releases Heavy Is The Crown | Song available; the 58th Floor is track 1 |
| Oct 30 | COLORS performance on YouTube | Format established; four-angle shoot documented |
| Oct 31 | Search spike on "58th Floor" | People discovering the song and the performance |
| Nov 2 | Templates released | First one-click way to recreate the format |
| Nov 3 | Top trending phrase across YouTube and TikTok | Five new domains registered same day |
| Nov 4-7 | Trend variants emerge | Hotel Lobby crossovers, duet combos, meme edits |
Each milestone had a source:
- The original COLORS performance is on YouTube
- First templates came from multiple platforms
- Album on Spotify
- Trend tracking came from Google Trends, YouTube and TikTok's trending sections
The spread was organic. No paid promotion, no celebrity pushback, no platform amplification beyond the algorithm doing its job.
Why it matters (and does not matter)
Why it matters: The 58th Floor proved that AI video works best when it lowers a barrier instead of replacing a skill. It did not replace cinematography or editing—it made it so people who had zero equipment could still participate in a specific trend. That is viral design.
It also proved that trend velocity tracks the inverse of effort required. The easier something is to make, the faster it spreads.
Why it does not matter: It is a trend. Trends peak and burn out. By next year, a different format will own the viral moment. The 58th Floor will be the thing people say "I remember that" about, the way they remember the CapCut dance templates or the Deep Dream meme era.
What lasts is the insight: AI video works when it is fun to make, instant to share, and low-stakes to participate in.
Frequently asked questions
How much does it cost to make a 58th Floor video?
The default clip is 10 credits at 720p, 12 seconds, 9:16. Credit packs start at $6.99 for 10 credits (about one video) and go up from there. No subscription required.
Can I use photos of my pet?
Yes. One clear photo of each animal, or one pet and one person. The format works as long as there is a clear face in each photo.
Do I have to log in to make a clip?
Yes. You need an account to upload photos and manage your credits.
How long does rendering take?
Most clips render in 2 to 5 minutes. You can close the tab while it renders and come back later to download the finished MP4.
Can I use the clip commercially?
The clip you make is for your personal and social media use. Using it to sell a product or service requires a separate commercial license. Check the pricing page for details.
Can I edit the video after download?
Yes. The download is a standard MP4 with audio baked in. You can trim it, add captions, crop it, or layer it into a larger edit—standard video editing applies.
Is the original song in the video?
Only if you upload your own audio. If you choose generated bars or the AI-written topic, the audio is produced by the model. No original Cordae or Anderson .Paak recording is included in any generated clip.
Why does everyone have the exact same set?
The 58th Floor performance had one specific layout: teal studio, hanging mic, four angles. Copying that exact setup is what makes it recognizable as the 58th Floor. If every video looked different, it would not be a trend anymore—it would just be AI video generation.
Can two people in different countries make a clip together?
Yes. The two photos do not need to be taken at the same time or in the same place. Upload one person's selfie from Tokyo and one person's from Toronto and the clip treats them as a duet.
What if my photos get rejected?
A rejected upload is usually a file format issue (check for JPG, PNG, or WEBP) or a size problem (under 20 MB). If a render fails, your credits are refunded automatically. Read the photo guide for the full list of what works.
Get started
Two photos in, a 12-second duet out. Open the 58th Floor AI video generator, or read the complete guide to the 58th Floor format if you want step-by-step photos and troubleshooting tips.