58th Floor AI Video Generator
One or two photos in, one teal-studio duet out. Your faces land on the two performers trading bars at a single hanging microphone, mouths synced to the vocal, in a landscape 16:9 video ready for YouTube, Reels or Shorts.
Reference performance
The 58th Floor preset is loaded by default. Remove it to use your own 3–15s motion reference.
First person
Required · left performer, or a photo containing both subjects.
Second person
Optional · right performer when using separate photos.
Use images of yourself, pets you own, or people who agreed to appear.
What the 58th Floor AI video gives you
The preset keeps the staging everyone recognises: a teal-lit studio, one microphone hanging between two performers, a wide two-shot with no cuts. Your photos control who is standing in it — identities, hair, outfits — while the room, the light, the mic and the blocking stay in place.
Output is landscape and follows the reference clip's length, 5–15 seconds. Faces are replaced on top of real performance motion, which is why the result moves like a performance instead of a photo pasted on a body.
The second photo is optional. Upload only the first person and the preset carries the pair, with your face on the left performer.
What is the 58th Floor AI trend?
"58th Floor" is the Anderson .Paak and Cordae track, performed as a duet on COLORS: a small teal-green room, one hanging microphone shared between two people, a wide shot with nowhere to hide. Both performances are genuinely different — one carries the verse, one answers on the hook — and both are moving the whole time, which is exactly what a face swap needs.
In late October 2026, AI versions of that staging spread across TikTok, Reels and X. Creators kept the room, the mic and the blocking, and replaced the two performers with friends, couples, coworkers, pets and fictional characters.
What this page generates is a recreation of the staging, not a re-upload of the original footage: the preset supplies the performance, your photos supply the people.
Three steps
How to make the 58th Floor AI video
Step 1
Upload one or two photos
One clear photo per performer, or a single shot containing both. Waist-up or full-body carries outfits better than a tight face crop; put the person who should take the verse in the first slot.
Step 2
Press generate
The 58th Floor performance is already loaded and the duration is taken from it. Nothing uploads until you sign in and press generate — choosing a file only creates a local preview.
Step 3
Add the sound and post
Download the MP4, upload it to TikTok, Reels or Shorts, then pick the licensed "58th Floor" sound from the platform's own music library rather than bundling the track into the file. Vertical feeds need a crop step first — see below.
Who people cast in the 58th Floor video
You and your best friend
The default, and the version that lands: give the verse to whoever talks more, the hook reads as a reaction.
A couple
The answering part has the better line, so it usually goes to whoever is likelier to have the last word.
Two coworkers
The shared microphone makes it read as a duet rather than two people in one photo — a Friday team-chat version writes itself.
Your parents
Formal outfits inside a teal studio read funnier than expected. One photo each, same light.
Your pets
One clear photo per pet, both facing the camera; the model needs the head angle to carry the lip movement.
Characters or mascots
Illustrated faces work when the eyes and mouth are readable. Anime stills with a side profile usually break the sync.
The wide shot problem: getting it into 9:16
The staging is a wide composition — two performers sitting toward the left and right thirds with the microphone hanging between them — which is the worst possible layout for a centre crop. Pull a straight 9:16 window out of the middle and you keep the mic and lose both people. Three ways to handle it, best-looking first:
- Crop with both performers in mind. Take a tall window that still holds both heads and the mic, even if side wall goes. Better than a dead-centre crop that halves both faces.
- Pan between them on the bar. Keep the wide frame and cut a moving tall window that follows whoever is rapping. Hides the crop, but needs a cut per line.
- Letterbox the wide export. Blur a copy behind a centred wide video, stacked into a tall canvas. Fast and safe, and clearly reads as a recycled wide clip.
Word-by-word captions belong in the empty space above the performers' heads — the one area a tall crop can safely cover.
Wide 16:9 studio framing next to a 9:16 crop that keeps both performers
15 seconds vs the 30-second extended cut
One render follows the reference you give it, 5–15 seconds. The standard version is a single 15-second pass that ends on the hook, so the last frame flows back into the first and viewers watch twice.
The 30-second cut going around is two renders cut together, not one long generation: run the verse from your photos, run the hook from the same photos, join them on the beat. Keep the left/right assignment identical across both — the lip sync is baked into each render, so a swapped assignment puts the wrong mouth on the join.
58th Floor AI vs Genjutsu
Both workflows can take motion from a source performance. The difference is scope: this page preloads the 58th Floor template and fixes the prompt, while the Genjutsu homepage is the general motion-transfer and object-swap workflow.
| This page | Genjutsu (home) | |
|---|---|---|
| What you upload | 1–2 photos; preset clip included | 4–30s source clip + references |
| Copies source motion | Yes, from the preset or your own clip | Yes, via Motion Transfer |
| Find the 58th Floor clip yourself | No | Yes, if that is the motion you want |
| Prompt writing | Not exposed | Optional |
| Default output | Matches the reference framing | Up to 1080p, source-driven |
Photo tips for a better swap
- One subject per photo when possible — crop a group shot to the person you mean.
- Both photos in similar light and from a similar angle: a daylight selfie next to a photo shot in a dark bar is the single most common reason a duo swap looks fake.
- Same crop for both, hairline to chin.
- No sunglasses, no hats pulled low, no heavy beauty filters.
- If the result looks off, replace the weaker photo before regenerating.
The audio question
Every version of this trend runs on the original commercial recording. For organic posts, the platform's music licence usually covers you — which is exactly why it is better to add the licensed sound inside TikTok or Instagram than to bake the track into the file. For ads, brand work or anything you are paid for, swap in a track you own: the mouth movement is generated against whatever audio you supply, so the format survives without the recording.
Free to set up, private by default
Load the preset and choose photos before spending anything; choosing a file only creates a local browser preview, and your images are not uploaded until you sign in and press generate. Generation uses your credit balance and the cost is shown before you run the clip; provider failures refund reserved credits automatically. Use images of yourself, pets you own, or people who agreed to appear.
58th Floor AI FAQ
What is the 58th Floor AI video?
A photo-to-video recreation of the teal-studio duet staging from Anderson .Paak and Cordae's COLORS performance. One or two photos go in, your faces come out, lips synced to the bar.
How do I make the 58th Floor AI video?
Upload one photo per performer, put the verse-taker in the first slot, and generate. The preset already holds the studio, the mic and the performance.
Do I need two photos or one?
Either. One photo with both subjects works, or upload each performer separately. The second slot is optional.
Can I do it with one person?
Yes. Upload the first person only and your face lands on the left performer while the preset carries the pair.
Do I need to find the original clip?
No. The preset supplies the performance. Upload your own 3–15s clip only if you have footage you have the rights to use.
Can I make it vertical for TikTok, Reels and Shorts?
The preset exports landscape. Crop keeping both performers in frame, pan between them on the bar, or letterbox the wide shot; a centre crop usually cuts a performer off.
How long is the clip?
It follows the reference, 5–15 seconds. A 30-second version is two renders cut together on the beat.
Is it free?
You can load the preset and choose photos before spending credits. Generation uses your credit balance and the cost is shown first.
Why does my version look fake?
Almost always the photo pair: mismatched lighting, a turned head, different crops. Replace the weaker photo before regenerating.
Can I use the original song?
For organic posts a platform's music licence usually covers you. For ads or paid work, use a track you own. Use images of yourself, pets you own, or people who agreed to appear.