How to Make a 58th Floor AI Video
Last updated:
To prepare a 58th Floor AI video with a friend, start with one portrait of each person. The performance reference is built in. This guide explains the steps and checks. Sign in, upload both portraits and use credits to generate a video.
1. Choose clear, separate portraits
Use one clear, evenly lit adult face per image. Assign photo 1 to the left performer and photo 2 to the right. Get permission from both people.
Use a front-facing or slightly turned portrait with a single face. Avoid sunglasses, hands across the face, strong beauty filters and tiny faces in distant group shots. A clear hairline and visible clothing help define the appearance. Keep each person in a separate file instead of a collage.
2. Assign left and right, then keep the reference
The built-in 58th Floor video supplies the movements, teal set, microphone and four-shot sequence. You do not need to upload a performance or write a prompt.
The right performer also appears alone in the close-up, so assign that portrait to photo 2. A text-only “two people rapping” request can produce a different performance. This workflow sends the source video with both identity images.
3. Understand the output and credits
15-second vertical MP4, 9:16, 480p by default, with the original soundtrack restored.
Credit pricing starts at $9.90 for 20 credits. One 15-second video uses 15 credits, so that pack covers 1 full video with 5 credits left. Prices are in USD; taxes may apply at checkout.
An account is needed to manage generation history and downloads. There is no free completed-video tier or verified fixed render time. You do not need a CapCut project, a local video editor or your own prompt.
4. Check the whole clip after each cut
The reference guides body movements, camera framing and cuts; the photos define the two identities. The tested result follows the main action sequence, but hands, expressions and timing can vary. Frame-for-frame motion or perfect lip sync is not guaranteed.
Check the opening duo, the right-person close-up and both later duo shots. Look for swapped identities, a missing person, distorted fingers or drifting mouth movements. Listen as well as watch. The sample was generated through the live website with automatic soundtrack restoration and no manual cut adjustments.
5. Download and share responsibly
Download the MP4 when the task completes and identify your shared clip as AI-generated. Videos are kept for 30 days, so save a local copy. Check the music rules of your posting platform; an attribution alone is not a license. Technical failures return the task’s credits. Cash-refund conditions are in the refund policy.
Who is taking the mic with you?
Prepare your two photos