Your photo is where the story starts
This is an effect preview. You can check progress and watch your finished video in My Videos.
Use this templateLoading…
58th Floor AI Video
Restoring draft…
- Estimated time
- 5–15 minutes
- Credit cost
- Loading…
Credits are automatically returned if generation fails due to a technical problem.
Your finished MP4 will be saved in My Videos, where you can play and download it.
Loading…
Use photos and likenesses you have permission to use.
Try More Viral Videos
What is 58th Floor AI?
Turn two portraits into a 15-second studio rap duet. Photo 1 opens on the left; Photo 2 takes over on the right, including the solo close-up. The performance combines AI-generated vocals with the template’s own backing track.
- What your photos control
- Photo 1 controls the left performer and the opening rap. Photo 2 controls the right performer, the following verse and solo close-up. Each photo guides its own character’s appearance and outfit throughout. The preview shows the supplied example result using these two portrait photos.
- Movement and background
- AI generates a new clip from the example’s routine and setting. Backgrounds, camera movement, expressions and occlusion may vary; frame-for-frame matching is not guaranteed.
Photo to video: a paired example
Compare the input photos with their corresponding video to assess the subject’s appearance and movement.
What happens in this video?
Three frames from this template’s showcase explain the scene and performance.
01 · Opening Start the duet
Photo 1 opens the rap on the left while Photo 2 keeps time with open-hand gestures, then takes over the verse.
02 · Middle Give the lead a close-up
The camera cuts closer to Photo 2’s character for expressive singing and hand gestures, keeping the same face and outfit.
03 · Later Show the whole stage
A wide shot reveals both people on a low stage before the video returns to a two-person close-up.
How to get a better result
Recommended photos
- Use two separate, sharp portraits with one person per image, even lighting and an unobstructed face.
- Show the shoulders, arms and outfit you want each person to wear. A full-body photo can give the wide stage shot more clothing detail.
- Even lighting and clear detail
- A subject outline distinct from the background
Try to avoid
- Collages, chat interfaces or watermarked screenshots
- Avoid group shots, strong beauty filters, dark sunglasses and a phone or hand covering the face.
- Avoid tight face-only crops or large objects blocking the torso. You do not need to photograph a microphone or recreate the studio.
- Blur, darkness or heavy filters
If the result needs improvement
- The appearance looks different
- Check the photo for the affected role: Photo 1 is left; Photo 2 is right and gets the solo close-up. Use a sharper face and visible clothing. The two identities are assigned independently, but a perfect match is not guaranteed.
- Limbs or props connect awkwardly
- Use photos with visible shoulders and arms, and a full-body view if possible. The prompt keeps each person’s clothing rather than copying the source actors’ outfits. Hands passing near the camera can still show distorted fingers or sleeves.
- Roles or positions get mixed up
- Photo 1 opens the rap on the left; Photo 2 takes over and gets the solo close-up. Both should appear in the wide shot. The model generates their voices with the picture, then the platform mixes in the original accompaniment. If one performer stays silent or lip movement is inaccurate, that is a generation issue; changing audio delay alone cannot fix it.
- The movement or performance differs
- This generates a new performance rather than replacing pixels in the source. The AI generates the picture and rap voices together; postprocessing keeps those voices and mixes in the original accompaniment. Voices differ from the source recording. Exact likeness and word-for-word lip-sync are not guaranteed.
- The task reports a generation failure
- Check the task status in My Videos first. Technical failures return deducted credits automatically; a successful video that does not meet personal expectations is a different situation. Each new generation costs credits, so review the refund policy before retrying.
How to Make a 58th Floor AI Video
- Step 01
Upload your photos
Upload two separate person photos, following the role order.
- Step 02
Generate your video
Choose 480p / 720p / 1080p, confirm the credits, and submit. The current output is 15 seconds in 9:16. You can leave the page while it processes, then check My Videos when ready.
- Step 03
Download and share
When ready, play and download the MP4 from My Videos and save a copy to your device.
My Videos
FAQ
Do I need two photos, or can I use a group picture?
Upload two separate photos, with one person in each. Use clear faces, even light and visible shoulders and clothing. A group picture does not assign both roles, and dark glasses or a phone covering the face can make the likeness less reliable.
Who is on the left, and who gets the close-up?
Photo 1 becomes the left performer and sings the opening phrases. Photo 2 takes over on the right and appears in the solo close-up. Both continue in the wide stage view and final two-person shot.
Will it keep our faces and clothes?
The prompt asks each person to keep the appearance, hairstyle, body shape, visible clothing and accessories from their own photo. It does not require the source actors’ hat, sunglasses, printed T-shirt, black jacket or jewelry. Details can still change during AI-generated movement.
Do we need to sing, and does it use our voices?
No singing recording is required. The AI generates the picture and both rap voices together, with Photo 1 opening and Photo 2 taking over. Postprocessing keeps the generated vocals and mixes in this template’s original accompaniment. It does not clone your voice or preserve the original singers’ recording. Exact lyrics and precise lip-sync can vary.
How many credits does 58th Floor AI cost?
You can watch the preview for free. Generating a 15-second video costs 50 credits at 480p, 100 at 720p, or 250 at 1080p. Check the selected quality and displayed credit cost before submitting.Pricing ↗
What length and format will I get?
The current output is an MP4 lasting 15 seconds in 9:16, with 480p / 720p / 1080p quality options. Check the current credit quote before generating.
What if a face changes or a hand looks wrong?
Try sharper single-person photos with clear faces, visible outfits and unobstructed arms. Hand gestures near the lens and turns can still distort details. A successful video that you choose to remake is a new paid task; technical failures automatically return the deducted credits.View refund policy ↗
Where do I get the finished video, and can I share it?
The finished MP4 appears in My Videos for preview and download, with no extra watermark added by the site. Make sure you have permission to use both people’s likenesses and the included audio before sharing.View refund policy ↗