Tiny Desk AI prompt: three you can copy
A good Tiny Desk AI prompt names the room (a cluttered office with shelves), the cast (lead at one chrome microphone, keys, two backing singers, sax), the order of the shots with seconds, the sound and what to leave out. The three prompts below were written for ByteDance’s Seedance 2.0 image-to-video model with your photo as @Image 1; other image-to-video tools need their own reference syntax. A text prompt can match the style of the original, not its exact timing.
Generate it without a promptThree prompts
SHOT SEQUENCE: 0-5 s medium-wide, handheld: @Image 1 sings into a chrome vintage microphone on a stand; a keys player at a red keyboard, two backing singers in headphones clap behind. 5-10 s close-up on @Image 1, who laughs mid-line and keeps singing. 10-13 s a saxophonist in headphones plays a short solo. 13-15 s back to the whole band, last chord, everyone laughs. STYLE: real, unedited concert footage in a small cluttered office, bookshelves, records, warm lamps, no stage. AUDIO: warm soul groove at about 92 BPM, electric piano, bass, light drums, handclaps; the lead sings in a natural chest voice. AVOID: subtitles, logos, the letters NPR, extra lead singers, a crowd or stage lights.
SHOT SEQUENCE: 0-3 s close-up of @Image 1 at a chrome microphone, eyes half closed, singing. 3-6 s they crack up on the chorus, shake their head and grin at someone off camera. 6-8 s they find the beat again and finish the line. STYLE: documentary handheld, shallow depth of field, shelves of books soft behind, warm practical light. AUDIO: the lead’s voice and the room, a soft keyboard underneath, a laugh breaking the line. AVOID: changing the face of @Image 1, text on screen, glasses or hats not in the photo.
SHOT SEQUENCE: 0-6 s close-up of @Image 1 in studio headphones playing a brass tenor saxophone into a microphone, eyes closed, swaying slightly; the camera drifts left. STYLE: the same cluttered office, bookshelves and sticky notes behind, warm lamp light, real footage. AUDIO: a bright, bluesy saxophone solo over electric piano and bass. AVOID: other people in frame, logos, text, a second instrument in the hands.
Common ways it goes wrong
- The face drifts: say that the face, hair and build come only from @Image 1 and stay the same in every shot.
- It turns into a stage show: name the office and forbid a crowd and stage lights.
- Two people sing lead: say there is one lead singer and one microphone.
- The singing looks like talking: ask for a natural chest voice, clearly singing, lips matching the words.
- Readable text appears: forbid subtitles, logos and the letters NPR.
Skip the prompt
Our generator does not use a text prompt like these: it keeps the viral clip’s shots, moves and laugh and puts your photo at the microphone, so you only add a photo. The trend’s sound is added to your download, with a silent copy.
Tiny Desk AI Prompt: questions
Which model are these Tiny Desk AI prompts for?
They were written for Seedance 2.0 image-to-video with reference images, where @Image 1 is your photo. Other models use different reference syntax, so replace @Image 1 with how your tool names an uploaded image.
Can a prompt recreate the original clip exactly?
No. A text prompt can match the look, the cast and the order of the shots, but not the exact timing, moves and sound of the 2023 performance. Copying that footage would also reuse NPR’s video.