Short answer
58th Floor AI is the photo-to-video version of Cordae and Anderson .Paak performing “58th Floor” on A COLORS SHOW. You upload one or two photos, and the studio returns a 15-second vertical clip of that teal-studio performance with the song underneath: the left seat finishes the hook, the right seat takes the verse. Nothing is filmed and no prompt is written, because the scene is one continuous take that already works.
Open the 58th Floor studioWhat is 58th Floor AI?
One teal wall, one hanging microphone, two people trading bars — and a seat for each of them.
Updated 2026-11-03
The performance everyone is copying
Cordae and Anderson .Paak recorded “58th Floor” for A COLORS SHOW, the German live-session series with a single microphone and no band on camera. The set is one solid teal wall, a low white platform, and a silver condenser microphone hanging from the ceiling on a cable. The clip is a single take: the camera holds a medium frame and drifts in slowly while the two of them trade lines.
The song’s title is the punchline of its own first verse — a man hiding in a closet working out that the window is a 58-storey drop — but the video does not dramatise it. There is no closet, no apartment, no window. Just two performers, a wall and a microphone, which is exactly why the image is so easy to copy.
Why a one-take clip became a template
Most trends people copy are hard to fake: they have cuts, camera moves, locations and props. This one has none of that. The frame never changes, the light never changes, and the audio carries the timing. That leaves exactly two decisions — who stands on the left and who stands on the right — and two decisions is the ideal shape for a photo-to-video product.
What the AI version actually does
The studio keeps the teal wall, the hanging microphone, the platform, the slow push-in and the audio, and replaces the two performers with the people in your photos. That is a per-shot identity replacement, not a face filter over the finished clip: the model re-renders the take with your faces, then the audio is laid back underneath.
You upload one photo or two. One photo takes the left seat and the original performer keeps the verse, which still reads as a duet. Two photos cast both seats and the clip becomes yours end to end.
58th Floor AI or Tiny Desk AI?
| 58th Floor AI | Tiny Desk AI | |
|---|---|---|
| Length | 15 seconds | 19 seconds |
| Seats | 2 (hook, verse) | 5 (lead, keys, two vocals, sax) |
| Set | Teal wall, one hanging mic | Record-store studio, red keyboard |
| Audio | “58th Floor” | “Outstanding” |
| Best for | Two people trading bars | A group, a band, a full cast |
Both are single-purpose tools rather than prompt boxes. If you want a band, use Tiny Desk AI; if you want the two-person performance, use 58th Floor AI.
How to make one
- Pick the photo that should carry the hook — that is the left seat and the face the clip opens on.
- Add a second photo for the verse, or leave it empty and let the original performer keep it.
- Choose 480p, 720p or 1080p, pay once, and download the 15-second vertical MP4 a few minutes later.