Do I need a prompt to make the Tiny Desk AI video?
No custom prompt is required in the Tiny Desk AI creator. Upload 1–5 photos, assign the band roles and choose the available quality and audio. The office-concert scene is built into the product. This guide is for understanding the scene or planning a creative brief for a general image-to-video workflow.
A prompt and a template do different jobs. A prompt is written scene direction; a built-in template supplies a prepared structure. If you want to create the video without managing your own model settings, use the browser workflow linked below.
What should an Outstanding Tiny Desk AI prompt describe?
Begin with five adult roles in a compact, warmly lit office corner: a lead seated at the microphone and keyboard, left and right backing singers, a laughing close-up performer and a saxophone player. Bookshelves, a desk and the instruments help communicate the scene without relying on a show logo.
Describe three shots in order: the lead and backing singers perform together; the fourth person gets a sustained laughing close-up; the saxophone player finishes the clip. Separate reference images for these shots make the cast assignment clearer than a single group image.
How should I connect the brief to my photos?
Use one clear photo per adult you want to include and assign each person a specific role. Make the lead photo easy to read: a sharp front-facing or chest-up portrait in even light. Separate photos can be taken in different places; a shared background is unnecessary.
Check who plays each of the five roles before creating the video. A written role list should distinguish the seated lead from the performer in the laughing close-up. Calling both people “the singer” can make the intended cast unclear.
Why should the audio choice be explicit?
A song title alone cannot supply a recording. The creator uses an actual motion reference and its audio track, rather than asking the video model to invent an Outstanding-style song.
Tiny Desk AI restores the original reference-clip soundtrack after rendering. Choose Original sound to include it in the MP4 or Silent to omit it.
What does a prompt leave up to the video model?
A creative brief communicates an intended scene, not exact measurements of the result. Likeness, expressions, hands, instruments and the placement of cuts can vary. Describing a laugh and a saxophone moment does not establish the exact timing or frame sequence a model will generate.
Review the whole output before posting. If a reference portrait is unclear, improving the source photo is a more useful next step than adding a long list of conflicting instructions. Keep the cast, visual action and audio direction easy to understand.
