58th Floor AI Video
RapTwins is a 58th Floor AI video generator that turns two photos into the COLORS teal-studio duet: photo 1 finishes the sung hook on the left, photo 2 raps the verse and gets the close-up, with the original song in the MP4. It is for two adult friends or a couple, from $9 a month (one video is 28–809 credits), 3 free credits at sign-up: 15 seconds, 9:16, 480p to 1080p, no watermark.
Updated
Make your 58th Floor AI video
1. Two photos: who sings, who raps
Which photos work best
JPG, PNG or WebP, 300–6000 px, up to 10 MB, one adult per photo. A sharp, front-facing waist-up photo with the mouth and eyes clearly visible, no sunglasses, and similar light in both photos keeps the two faces apart through the close-up.
2. Model and quality
480p · 15 s · 9:16 vertical MP4 with the original song (explicit vocals), in sync with the hand-off
Add a photo for the hook and a photo for the verse.
New accounts get 3 free sign-up credits right now, fewer than one video costs (from 28 credits).
A finished duet: 15 s with the song · 9:16 · your video keeps this teal studio, the hanging mic, the camera cuts and this recording, with your two faces and outfits.
58th Floor AI video examples
Two finished duets in the same teal studio with the hanging microphone and the same song, and eight seconds of the original COLORS performance. The first is the exact 15-second cut your video follows; the second is the same hand-off framed a little closer. Tap a clip for sound.
The exact 15 s cut: pink suit and black tank top
The same 15 seconds the generator makes, with the song: a man in a pink suit and sunglasses finishes the hook on the left, the man in a black tank top raps the verse and gets the close-up, then both are seen head to toe before they return to the hanging mic. 15 seconds, 9:16, with sound.
15 s · with sound · vertical
Same hand-off, closer framing: silver jacket and black shirt
The same sung hook, hand-off and teal studio with the hanging microphone, framed a little closer and with slightly different gestures: a blond performer in a silver-and-green jacket on the left, a rapper in a black shirt on the right. 15 seconds, 9:16, with sound.
15 s · with sound · vertical
The original COLORS performance: the first 8 seconds
Eight seconds of the A COLORS SHOW performance that the trend copies: Anderson .Paak finishes the hook on the left, then Cordae takes the verse into the close-up. Your video follows the full 15 seconds of this exchange. 8 seconds, 9:16, with sound.
8 s · with sound · vertical
How do you make a 58th Floor AI video from two photos?
Add one photo for the singer and one for the rapper, pick a model and a resolution, tick the permission box and press the button. Both photos and the studio guide go to the model in one generation, then the original song is laid under the result. It usually takes 3–8 minutes, you can leave the page while it runs, and the price shows before you start.
1Add photo 1 and photo 2
Photo 1 sings the end of the hook on the left; photo 2 raps the verse on the right and gets the close-up. Use Swap hook and verse if you put them the wrong way round.
2Pick a model and resolution
Seedance 2.0 Mini at 480p is the default (28 credits); Seedance 2.5 gives the steadiest faces, up to 1080p.
3Confirm both people agreed
Tick the box: both photos show adults who said yes to being in the video.
4Download the MP4
A 15-second 9:16 video with the song in sync and no watermark, ready for TikTok, Reels or Shorts.
In the 58th Floor video, which photo sings the hook and which raps the verse?
Photo 1 is the person on the viewer’s left who sings the last line of the hook and then turns to listen; photo 2 is the person on the right who takes the verse, gets the solo close-up and keeps rapping to the end. Put the friend who should rap in photo 2. The roles never change mid-video, and both original performers are replaced.
| Seconds | Shot | Photo 1 (left) | Photo 2 (right) |
|---|---|---|---|
| 0–5.5 s | Both at the hanging mic | Sings the last line of the hook, then turns to listen | Waits, then starts the verse |
| 5.5–8.7 s | Close-up on the right | Out of frame or partly hidden | Raps the verse to camera |
| 8.7–11.5 s | Both, head to toe on the platform | Reacts without mouthing the verse | Keeps rapping |
| 11.5–15 s | Back to the shared mic | Listens and moves to the beat | Finishes the verse to the final pose |
Only one silver microphone hangs between the two performers, on its cable, for the whole video; nobody holds it. There is no mode with three people, and the generator never adds a third performer.
Why do the faces swap in a 58th Floor AI video, and how do you stop it?
Faces swap when the model can’t tell the two people apart after a camera cut: similar lighting, the same hair or two small, blurry selfies make it guess. Two sharp, front-facing photos with similar light, the rapper in photo 2, and Seedance 2.5 or a higher resolution fix most of it. The close-up is always photo 2.
| What you see | Why it happens | What to change |
|---|---|---|
| The two faces trade places halfway through | The photos look alike in lighting, framing or hair, so the model loses track after a cut | Use two photos with similar light and size, faces clearly different, and put the rapper in photo 2 |
| The close-up shows the wrong person | The close-up always belongs to photo 2 | Press Swap hook and verse before you start, or upload in the other order |
| A face drifts in the close-up | A small, dark or blurry photo has too little detail for a full-screen face | A sharp, front-facing waist-up photo; 720p or 1080p; Seedance 2.5 or 2.0 Pro |
| Sunglasses or a hat appear from nowhere | Covered eyes or a cropped head leave the model guessing | A photo with the eyes and hairline visible, no sunglasses |
| Both mouths move at the same time | The listener’s mouth is hidden or the photo is a side view | Front-facing photos with the mouth closed or relaxed; a waist-up photo for the listener |
| Clothes change between shots | A head-only photo forces the model to invent the outfit | A waist-up or full-body photo shows it what to keep |
How long is a 58th Floor AI video, and does it come with the song?
Every video is 15 seconds, vertical 9:16, with the original 58th Floor recording laid under the picture after it is made, so the hook, the hand-off and the verse land where they should. You don’t upload music or pick a sound. The recording has explicit vocals, and there is no clean version in the generator.
| Your video | |
|---|---|
| Length | 15 seconds, the whole hook-to-verse exchange |
| Frame | 9:16 vertical |
| Resolution | 480p, 720p or 1080p by model; 768P on MiniMax H3 |
| Sound | The original song, explicit vocals |
| Watermark | None |
| File | One MP4 |
For a square or landscape post, crop in your editor and keep both performers and the mic inside the frame; the cut itself is only made in 9:16.
Which model should you pick for 58th Floor?
Start with Seedance 2.0 Mini at 480p (28 credits) to check your photos, then pick Seedance 2.5 for the steadiest faces through the close-up or Seedance 2.0 Pro for 1080p without the photo review. All five models get the same two photos, the same studio guide, the same instruction and the same song afterwards.
| Model | Resolution | Credits | Good for |
|---|---|---|---|
| Seedance 2.0 Mini | 480p, 720p | 28–60 credits | The default and the cheapest; a first try |
| Seedance 2.0 Fast | 480p, 720p | 60–130 credits | A step up from Mini at the same tiers |
| Seedance 2.0 Pro | 480p, 720p, 1080p | 102–545 credits | Sharper faces up to 1080p without the safety review |
| Seedance 2.5 | 480p, 720p, 1080p | 146–809 credits | The steadiest faces through the close-up; a quick photo review first |
| MiniMax H3 | 768P | 145 credits | A different look at 768P, one fixed price |
How much does a 58th Floor AI video cost?
One 15-second duet costs 28 to 809 credits, charged once when it starts and refunded automatically if it fails. The cheapest is Seedance 2.0 Mini at 480p for 28 credits (about $1.04); Seedance 2.0 Mini at 720p is 60, MiniMax H3 is 145, and Seedance 2.5 is 146 at 480p up to 809 at 1080p. The Starter plan ($9 a month, 240 credits) covers 8 videos at the default setting. The one-time $5 trial pack (73 credits) covers 2 videos on Seedance 2.0 Mini at 480p.
| Model | Length | Resolution | Credits | About |
|---|---|---|---|---|
| Seedance 2.0 Mini | 15 s | 480p | 28 credits | $1.04 |
| Seedance 2.0 Mini | 15 s | 720p | 60 credits | $2.23 |
| Seedance 2.0 Fast | 15 s | 480p | 60 credits | $2.23 |
| Seedance 2.0 Fast | 15 s | 720p | 130 credits | $4.83 |
| Seedance 2.0 Pro | 15 s | 480p | 102 credits | $3.79 |
| Seedance 2.0 Pro | 15 s | 720p | 217 credits | $8.07 |
| Seedance 2.0 Pro | 15 s | 1080p | 545 credits | $20.26 |
| Seedance 2.5 | 15 s | 480p | 146 credits | $5.43 |
| Seedance 2.5 | 15 s | 720p | 328 credits | $12.19 |
| Seedance 2.5 | 15 s | 1080p | 809 credits | $30.08 |
| MiniMax H3 | 15 s | 768P | 145 credits | $5.39 |
Is there a free 58th Floor AI video, and is there a watermark?
Not fully: a new account gets 3 sign-up credits, fewer than one video costs (from 28 credits, about $1.04), so the rest comes from a plan or credit pack. Every download has the song and no watermark, and a failed video gives the credits back automatically. See plans and credit packs.
| Sign-up credits | Plan or pack credits | |
|---|---|---|
| 58th Floor AI video | 3 credits, not enough for one | 28–809 credits each |
| Watermark | None | None |
| Download | MP4 with the song | MP4 with the song |
| Failed video | Credits back automatically | Credits back automatically |
How does it compare with Starrd, Videosi and AI Meme Videos?
All four take two photos and give back a short video with the song. They differ in length, frame and how the price is set: Starrd charges one fixed price per video, Videosi makes 13 seconds in several frames, AI Meme Videos and this page price by model and resolution.
| Photos | Price per video | Free to start | Watermark | Output | |
|---|---|---|---|---|---|
| Starrd | Two | 150 credits; packs from $4.99 for 100 | 20-credit welcome bonus, not enough for one | Not published | 15 s video; a separate 30 s Extended template |
| Videosi | Two | 11 credits (Standard, 720p); packs from $6.99 for 10 | Not free | None | 13 s MP4 in 9:16, 1:1 or 16:9, with the sound |
| AI Meme Videos | Two | 10 credits (Standard, 720p), 5 at 480p; single pack $6.99 for 10 | Not published | None on paid videos | 15 s video |
| RapTwins | Two, hook and verse | 28–809 credits | 3 credits, not enough for one | None | 15 s 9:16, 480p–1080p, with the song |
What can’t this 58th Floor AI generator do?
It can’t change the song, the lyrics, the studio, the cuts or the length: the 15-second cut and its recording are fixed, which is why the hand-off stays on the beat. It needs two photos of two adults, makes 9:16 only, has no 30-second extended version and does not animate pets. A face can blur for a frame or two in the close-up.
No extended version
Every video is the 15-second hook-to-verse exchange; there is no 30-second cut.
People only
The guide and the instruction are written for two adult performers; dogs, cats and children are not supported.
No other music or lyrics
The file always carries the original recording; to post with another sound, add it in the app and turn the original audio down.
No edits after
A new take is a new video at the same price, so check both photos and their order first.
What is the 58th Floor trend from A COLORS SHOW?
58th Floor is a song by Anderson .Paak and Cordae from their joint album Heavy Is the Crown, performed on A COLORS SHOW in a plain teal studio with one vintage microphone hanging between them: .Paak sings the hook and Cordae takes the verse. The 58th Floor COLORS AI trend puts two friends into that performance.
COLORS posted the performance on October 30, 2026, and AI versions with two friends’ faces followed within days. COLORS’ note on the 58th Floor performance is the original. This generator follows 15 seconds of it with two other people; it is not affiliated with COLORS, Anderson .Paak or Cordae.
58th Floor AI video FAQ
Do we need two separate photos, or can one photo of both of us work?
Two separate photos, one person in each. If you only have a picture of the two of you, crop it into two photos, one face each, so the model knows who sings and who raps.
Is there an official TikTok filter or CapCut template for 58th Floor?
We know of none as of November 4, 2026. People make the trend in AI video tools like this one and then upload it to TikTok, Reels or Shorts as an MP4.
Does the 58th Floor song have explicit lyrics?
Yes. The recording in the video has explicit vocals and there is no clean version in the generator. If you need a family-friendly post, add a different sound in the app and turn the original audio down.
Can I make the 30-second 58th Floor extended version?
No. Every video here is the 15-second hook-to-verse cut, which keeps the hand-off and the close-up in time with the song.
Can I use photos of my dogs or cats for the 58th Floor AI video?
No. This generator is made for two adult people; pet photos are not supported and may come out as people or fail.
Can I change the lyrics or write my own prompt?
No. The instruction, the studio guide and the song are fixed so every video matches the original cut. Your choices are the two photos, their order, the model and the resolution.
What happens if the video fails?
The credits go back automatically and nothing is downloaded. A finished video you would like to redo is a new video at the same price.
Check this answer with an AI assistant
“How do I make a 58th Floor AI video with two photos?”
Your 58th Floor duet is two photos away.
Make our 58th Floor duet