Skip to content

AI Tiny Desk Video Generator

Your clip. Your cast. A new take on the performance.

Add a clip to estimate credits

Review the cost before you generate.

Desk-concert reference clips

Full supplied reference clip (about 19 seconds), not an AI Tiny Desk result. Recreate loads this source video; add your own reference images.

The first 15 seconds of a supplied external reference, not an AI Tiny Desk result. Recreate loads this source video; add your own reference images.

External reference: the first 10 seconds of College football ensemble. This is an excerpt of the same clip, not a separate AI Tiny Desk result. Recreate loads the excerpt; add your own reference images.

Make a desk-concert edit with a cast of your own. Start with a performance video you have permission to use, add reference images, and describe the change. AI Tiny Desk lets you work from the movement in your clip with Motion Transfer, or request a focused replacement with Object Swap.

What is the Tiny Desk AI trend?

Tiny Desk AI videos recast a music performance with a different lineup. An October 2026 example puts college football coaches into the scene. Sport & State's October 22 creator interview identifies its maker as Jake Whipple, posting as The Burner. The familiar performance and unexpected cast are the core of the joke.

The "Outstanding" connection comes from Charlie Wilson's performance of the Gap Band song at NPR's Tiny Desk. You can watch the original on NPR Music's official channel. The official upload is linked for background. The separate reference clips shown here were supplied from outside AI Tiny Desk and are not results generated with this tool.

AI Tiny Desk is an independent video tool. Bring your own permitted footage and images to make a version around your idea. The workspace edits an existing video; a photo alone does not supply the performance. We are not affiliated with NPR, NPR Music, Tiny Desk Concerts, Charlie Wilson, any other artist, university or sports team.

These supplied clips are external references, not results generated with AI Tiny Desk. Recreate loads a source video; add your own reference images before generating. The 10-second ensemble clip is an excerpt of the longer reference.

Two ways to reinterpret your performance

Motion Transfer: use the movement as your guide

10-second reference excerpt
These supplied clips are external references, not results generated with AI Tiny Desk. Recreate loads a source video; add your own reference images before generating. The 10-second ensemble clip is an excerpt of the longer reference.

Choose Motion Transfer when the gestures and camera movement in your source are the starting point. Add an image of the character you want to see, then describe the appearance and setting. A short, steady performance gives you a clear sequence to work with.

Start with one visible performer and one reference image. If your idea includes more people, identify each intended role in the prompt and match it to the correct image number. You can add up to eight reference images; those image numbers are instructions for the model, not locked face assignments.

Watch for changes in faces, clothing and movement throughout the result. Hands crossing a face, overlapping performers and camera cuts can make a change harder to judge. Begin with a short selection before committing credits to a longer version.

Object Swap: request one specific change

Choose Object Swap for a focused replacement within the shot. You might change the main performer, an outfit or a visible prop. Identify the target by its position, appearance and action, and explain which parts of the scene should stay consistent.

For a first attempt, select a passage where the target stays clearly visible. Add a reference image that shows the details you want. A request to change one performer is easier to review than a prompt that also changes every other person and the entire room.

Inspect nearby faces, microphones, hands and background details after generation. Object Swap can affect areas you asked to preserve. It also needs at least 409,600 pixels in each prepared source frame; heavy cropping can take a video below that limit.

Playback and checklist diagram for reviewing faces, hands, timing and audio
Review your result
Your ideaWhat to prepareSuggested approach
A new character following the performanceA clear source clip and character referenceMotion Transfer
One performer, outfit or prop replacedA reference image and a precise target descriptionObject Swap
A different song or tighter musical timingYour downloaded result and audio you may useFinish in a separate video editor

How to make your video in three steps

Choose a source clip Add reference images Review your result

  1. 1. Add your performance clip

    Workflow diagram with two performers and a highlighted section of a video timeline
    Choose a source clip

    Choose Recreate on a reference card to load its source video, or choose Add video, then Upload my video to use your own clip. Use Edit clip to select 4–30 seconds and adjust the crop. Pick a passage with clear faces, visible gestures and enough room around the person you want to change.

    Use footage you have permission to edit and upload. Loading a reference does not include character images or start generation. Keep only the audio you are entitled to use, and preview the full selection before saving it.

  2. 2. Add images and describe the change

    Four numbered character cards illustrating reference-image order
    Add reference images

    Add one to eight reference images. A clear face and visible outfit help you explain a character change. Choose Motion Transfer or Object Swap, then edit the prompt to match your source, image order and intended result.

    The starter prompt uses Reference 1. If you add more images, explain their roles; if you reorder them, update the numbers in your prompt. Include what should stay consistent, such as the camera angle, microphone placement or other performers.

  3. 3. Check the quote, then generate

    Playback and checklist diagram for reviewing faces, hands, timing and audio
    Review your result

    Choose 480p or 720p and select Generate. Sign in if prompted, review the confirmed credit quote and add credits if needed. Generation starts when you choose Confirm & generate. Signing in or buying credits does not start it automatically.

    When the result is ready, watch the whole clip and listen to its audio. Check faces, hands and timing before downloading. Your account holds completed results for a limited period, so save the video you want to keep within 30 days.

These diagrams illustrate choosing a clip, adding reference images and reviewing a result. They are not screenshots of completed generations or a before-and-after comparison.

Prompts for your desk-concert edit

The Motion Transfer prompt below is the starting text in the workspace. Adjust it to describe your actual clip. Switching modes keeps your text, so copy or adapt the Object Swap example when you change modes. These prompts describe an intended result; they are not tested guarantees.

Four numbered character cards illustrating reference-image order
Add reference images

Motion Transfer: a new lead performer

Use the adult character in Reference 1 as the main performer in this source video. Use the source performance as guidance for the gestures, posture, camera movement and timing. Match the reference character's face, hairstyle and outfit. Keep the staging and the other performers as consistent with the source as possible. Keep hands and microphones distinct. Do not add new people, logos, captions or camera cuts.

Use a source with a clearly identifiable main performer. For a group, describe each role and its reference image explicitly; separate identities are not guaranteed.

Object Swap: replace the main performer

Replace only the main adult performer nearest the camera with the adult character in Reference 1. Match the reference face, hair and outfit to the source pose, scale and lighting. Keep the other people, their clothing, the microphones, instruments, background and camera framing as consistent as possible. Follow the replacement throughout the selected clip. Do not add people, logos or text.

Rewrite the target description if the person you want to change is not nearest the camera. Review the entire result for changes to other people or objects.

Keep faces, gestures and microphones in view

Diagram comparing a wide group shot with a narrow portrait crop

Choose the performance moment before choosing the crop. A hand gesture, a glance at another performer or a shared movement may be the detail that makes the edit work. Leave space for it throughout the clip, including the first and last frames.

A tight vertical crop can remove people from a wide group shot. Preview the whole selection before saving, and avoid cutting through faces or microphones. For Object Swap, the prepared frame must retain at least 409,600 pixels; 854 × 480 is one example that meets the requirement.

Audio needs its own review. The clip editor can keep or remove the source audio, but the generated result may not preserve it or match every mouth movement. Listen before sharing, and use a separate editor for any final soundtrack or timing adjustments.

Credits for your next take

Buy a one-time credit pack for Motion Transfer or Object Swap. Your clip length and resolution determine the cost, which you review before generating.

Trial

—

one-time

480Credits

480p · up to 8s total

720p needs at least 520 credits

Mini

—

one-time

1,000Credits

480p · up to 16s total

720p · up to 7s total

Starter

—

one-time

1,500Credits

480p · up to 25s total

720p · up to 11s total

Creator

—

one-time

3,300Credits

480p · up to 55s total

720p · up to 25s total

Included with every pack
  • Motion Transfer and Object Swap
  • Private history and MP4 downloads
  • No credit expiry while the service is active; no auto-renewal

A 10-second clip uses 600 credits at 480p or 1,300 at 720p.

480p: 60 credits/sec · 720p: 130 credits/sec

Pack allowances are alternative uses of one balance and can cover multiple clips. Each clip must be 4–30 seconds. Combine packs for longer clips or 720p.

Pricing and billing details

Check the cost of your selected clip

One-time credit packs. Both modes use the same rates.

Generation uses 60 credits per billed second at 480p or 130 at 720p. The prepared clip duration rounds up to the next whole second: a 4.2-second selection is billed as five seconds. Review the quote after your final trim and resolution choice.

Video duration480p720p
4 seconds240 credits520 credits
5 seconds300 credits650 credits
10 seconds600 credits1,300 credits
16 seconds960 credits2,080 credits

The lowest-priced pack includes 480 base credits for —. Any eligible bonus is shown separately. Compare your available balance with the confirmed quote; one pack may not cover your full clip or chosen resolution. See credit packs and billing details.

Tiny Desk AI questions, answered

The name points to edits of Charlie Wilson performing the Gap Band song Outstanding at NPR's Tiny Desk. The edits change the people on screen. The official original is linked above for context. The supplied reference clips are external material, not AI Tiny Desk results, and their inclusion does not grant permission to reuse the original performance or soundtrack.

Give your performance a new cast

Bring a clip and reference images you have permission to use. Start with a short scene, describe your idea and review the credit quote before generating.

Start my video