How do I recreate the Outstanding Tiny Desk AI video with my friends' pictures?

Collect one clear, front-facing photo per friend, assign each to a band seat (lead, keys, two backup singers, sax), and render the template. With a generator it takes three steps; by hand you need character references, motion references and a video model.

Can I use one group photo for everyone?

It works poorly. Faces in a group shot are small and get mixed up. Use a separate portrait for each person.

Which AI video model is used for the Tiny Desk trend?

Most creators use a reference-to-video model such as Seedance, which takes both character images and a motion clip. Plain image-to-video models invent a new performance instead of copying the original one.

Do my friends need to agree?

Yes. Only use photos of adults who agreed to be in the video, and label the result as AI when you post it.

How to recreate the Outstanding Tiny Desk AI video with your friends' pictures

Updated October 29, 2026 · Tiny Desk AI team

To recreate the Outstanding Tiny Desk AI video, you need one clear photo per friend, a seat for each of them in the band, and a model that copies the original performance's motion. There are two ways to do it: three steps with a generator, or by hand with your own video model.

What you need

1–5 photos, one person per photo, face toward the camera
A plan for who plays which part: lead singer, keys, two backup singers, sax
Their permission to put them in the video

Option 1: the 3-step way

Upload your friends' pictures. On the Tiny Desk AI generator, add one photo to each seat. Only the lead singer is required.
Pick a template. The full band template matches the original; the one-photo template puts one person in every seat.
Render and download. After 2–5 minutes you get a 15-second 9:16 video with sound.

Option 2: the DIY way

If you already pay for a video model, you can build it yourself. It takes longer, but you control every shot.

1. Make a character sheet per friend

Turn each photo into a clean reference: the face from the front and three-quarter view, plus the outfit they will wear. Decide wardrobe now (the red hat goes to the lead).

2. Make motion references

The performance has four beats: the group chorus, the lead close-up, a laugh, and the sax solo. Create a short, simple motion clip for each beat that you have the rights to use, such as a sketch-style black-and-white animation of the poses and camera moves.

3. Generate each clip

Use a reference-to-video model (Seedance is the common choice). Upload the character sheets for the people in that shot plus its motion clip. Plain image-to-video invents a new performance, so it won't match. Our Tiny Desk AI prompts page has a prompt for every shot.

4. Edit them together

Place the four clips on a 9:16 timeline, cut on the beat, and trim instead of speeding clips up. Add the trending sound in the app when you post.

Photo tips that matter most

Use a separate portrait for each person, not a cropped group shot.
Front-facing, even light, no sunglasses or hands over the face.
Give the lead singer seat your sharpest photo: it gets the close-up.

Common problems

Faces swap between people: photos are too similar in angle and light, or a group photo was used.
Mouths drift off the beat: shorten the clip and regenerate that shot only.
Hands on the sax look wrong: video models still struggle with fast fingers; a closer crop on the face hides it.

Want the backstory first? Read about the Outstanding Tiny Desk AI meme.