Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
5
5
moment
stringclasses
6 values
framing
stringclasses
5 values
aspect_ratio
stringclasses
2 values
prompt
stringlengths
136
178
visual_check
stringlengths
44
61
td-01
opening
wide
16:9
A fictional acoustic quartet starts a tune in a book-lined rehearsal room. Hold a wide view as the vocalist takes a breath and the drummer taps the first beat with brushes.
All performers stay visible until the first phrase begins.
td-02
opening
medium
9:16
Frame a singer beside a small desk lamp in a narrow rehearsal nook. Let a bassist in the background nod once to cue the opening. Keep the camera still.
The singer and lamp stay inside the portrait crop.
td-03
verse
medium
16:9
An indie folk vocalist sings a quiet verse while a seated guitarist plays beside a shelf of paperbacks. Ease the camera forward slightly without changing the seating arrangement.
The two performers stay on their original sides of the frame.
td-04
verse
close-up
9:16
Hold on a fictional jazz singer from the shoulders up. A warm practical light catches one cheek as the singer pauses between lines and looks toward the band.
Face detail stays steady through the glance.
td-05
chorus
two-shot
16:9
Two fictional vocalists lean toward separate microphones for a brief call-and-response. Keep both mouths visible and let the listener react only after the other finishes.
The two vocal parts read as distinct actions.
td-06
chorus
medium
9:16
Keep the lead performer centered while a second singer enters the edge of the portrait frame for one harmony phrase. Use a stationary camera and soft daylight.
The second singer enters without obscuring the lead.
td-07
instrumental
insert
16:9
Show a short insert of a percussionist brushing a snare in a cramped music room. Keep the hands and rim in view and use shallow but stable focus.
Both brushes remain visible and do not merge with the drum.
td-08
instrumental
medium
9:16
A fictional keyboard player answers the previous vocal line with a short riff. Frame from waist to forehead so the keys and a quick smile fit in the same shot.
The playing hands and face remain in frame together.
td-09
transition
wide
16:9
In a small listening room a singer turns toward a cellist during a quiet transition. Use one continuous wide shot while the other musicians wait for their cue.
No sudden cuts or changes in performer placement.
td-10
transition
two-shot
9:16
A singer lowers the microphone and exchanges a small nod with a guitarist. Keep the two faces on different depth planes and use a fixed camera.
Both expressions are readable before the next phrase.
td-11
ending
wide
16:9
The fictional band lands on a final chord and lets it ring. Hold the wide frame for a moment as shoulders relax and the vocalist smiles.
Hold the final composition through the visible release.
td-12
ending
medium
9:16
A fictional singer finishes the last line and steps half a pace away from the desk microphone. Let the camera hold on the performer as the music ends.
The microphone and face both stay within the portrait crop.

Outstanding Tiny Desk AI Prompt Examples

Twelve original, synthetic English prompts for planning short, fictional desk-concert videos. The examples span six moments in a performance and cover landscape and portrait framing. Use them as starting points for storyboards or to compare how video generators handle framing, performer continuity and hand movement.

Data

Download prompts.csv to adapt the examples to your workflow.

Field Meaning
id Stable example identifier
moment Opening, verse, chorus, instrumental, transition or ending
framing Suggested shot size
aspect_ratio Landscape (16:9) or portrait (9:16)
prompt Original English shot description
visual_check Observable detail to inspect in a generated clip

For a framing comparison, run a landscape example and a portrait example for the same performance moment. Keep the model and reference image constant, then inspect the listed visual checks. The prompts describe different shot setups, so this exercise compares the overall framing choices.

Provenance and scope

These text-only examples were written from scratch for this dataset. They contain fictional scenes and generic roles. The dataset includes no ViGen production prompts, source footage, recordings, reference images or user data. The visual checks are review criteria. No generated videos or measured results are included.

Related tool

ViGen's Tiny Desk AI Video Generator is a public tool for creating a fictional desk-concert video from photos. This dataset is independent of NPR and the Tiny Desk concert series.

Downloads last month
39

Collection including aiddroid/outstanding-tiny-desk-ai-prompt-examples