Tiny Desk AI prompts: how to prompt the Outstanding meme
Example Tiny Desk AI prompts for each shot of the Outstanding meme, why prompting five faces is hard, and how to skip prompts entirely with one photo per seat.
Why one prompt is not enough
The Outstanding Tiny Desk meme is five cuts, up to five different people and a song that has to line up with the mouths. A general AI video tool makes one shot at a time, so you need a prompt per shot, a reference photo per face and an editor to put it together.
A prompt for each shot
Start from these and swap in your own description. Attach the person’s photo as the reference image for each shot:
- Lead: “Vertical 9:16 handheld concert shot. [Person] sings into a vintage microphone behind a cluttered office desk, bookshelves and posters behind, warm lamp light, joyful.”
- Band: “Wide 9:16 shot of a small band squeezed behind an office desk: someone at a red keyboard in front, two backing singers swaying behind, warm light.”
- Laugh: “Tight close-up of [person] bursting into a huge laugh mid-song, eyes closed, warm light, shallow focus.”
- Sax: “Medium shot of [person] playing a saxophone solo between bookshelves, eyes closed, leaning into it.”
- Ending: “Wide shot of the whole band behind the desk finishing the song together, everyone smiling.”
Prompting tips that help
What usually makes the difference:
- Say “vertical 9:16” in every prompt so the shots match
- Repeat the same set description (desk, bookshelves, warm lamp light) word for word
- Use one clear, front-facing photo per person as the reference
- Keep each shot short: three to five seconds
- Add the song and the cuts in an editor afterwards
Where prompting goes wrong
Faces drift between shots, the band changes from cut to cut, lip movement doesn’t follow the song, and five people means five reference photos to juggle. Expect several tries per shot.
Skip the prompts
Tiny Desk AI has the shots, the set and the song built in. Add one photo per seat (up to five), check the credit total and get an ~18s 9:16 video with sound. No prompt, no editor.