Tiny Desk AI Video Generator

Tiny Desk AI turns portraits into a desk-side concert video with a singer, a small band and shelves of books behind them. The Outstanding variation uses an unexpected cast in a familiar performance setting. On this page, you can replace up to five performers using separate photos.

Keep or replace the five default photos, choose 720p or 1080p and check the credits shown on Generate. Sign in and submit your video, then follow its progress in History. The performance reference, role assignments and soundtrack setting are already prepared.

One new selfie is enough to personalize a role. For example, replace the lead singer and keep the other four default photos. You do not need to upload five new images, but every role must contain either a default photo or your replacement before generation.

Yes. The roles are left singer, center singer, right singer, lead singer and saxophone player. Assign one portrait to each role you want to change. Replace one, several or all five photos; the template keeps a separate identity assignment for each performer.

Use a clear, front-facing portrait of one person with the head, shoulders and some clothing visible. JPG or PNG is recommended; the photo picker also accepts WebP and converts it to JPEG before upload. Each photo must be under 10 MB. Both dimensions must be between 300 and 3840 pixels, and the shorter side must not exceed 2160 pixels. The included default photos are 480 × 640 JPEGs.

No. This template already includes the concert reference and instructions for assigning the five portraits to their roles. Change the photos and output quality, then generate. A separate video editor is only needed if you want to add your own captions, cuts or other finishing touches afterward.

This page uses Kling 4.0 with the five portrait references and a fixed concert video reference. The model is preset in the creation panel, so you do not need to choose it yourself.

The template requests the concert reference audio together with the generated picture. It does not record or clone your voice from a selfie. Listen to the finished video and check the mouth movements, faces and instruments before sharing; soundtrack and performance accuracy can vary.

The creation panel requests a 19-second video in a fixed 9:16 vertical frame. Choose 720p or 1080p for the output resolution. The examples elsewhere on this page may be shorter excerpts; they do not change the duration requested by this template.

You can watch the examples and replace the default photos without purchasing a credit pack. Generating a new video requires sign-in and enough account credits. There is no separate free generated preview offered by this template.

The Generate button shows the credit quote for your selected resolution before you submit. The quote covers the fixed performance template with up to five replacement portraits; changing more band photos does not add a per-person charge. Use your existing balance or buy a credit pack from the pricing section.

Wait time varies with the queue and processing. Follow the task in History; the percentage is an estimate rather than a guaranteed completion time. When the video finishes, open its card to review and download it. Use Refresh if you want to check for an updated task status.

Read the error shown in the preview area before retrying. For an image error, check the actual file format, dimensions and size, then replace the affected photo. A failed attempt does not turn an older video into its result. You can select View previous videos to return to earlier work; contact support if a task or credit issue remains unresolved.

Yes. Open the page in your mobile browser and tap Create Video to open the editor. Replace the photos, generate and review the result in History. Download the finished vertical video from its work card without installing a separate editing app.

Use photos of yourself or people who have agreed to the edit. Make it clear that the performance is AI-generated when the context could mislead viewers. Before publishing or using the video commercially, check that you have permission for any faces, music and footage involved; generating a clip does not grant those rights.