Start with one clear portrait for a solo, or two separate portraits for a duet. The first person is the lead singer; the second becomes the saxophone partner. You do not need to film yourself or write a prompt.
The current studio requests 15-second, 720p office-concert videos with original AI audio. This is a new performance inspired by the meme format, not an edit of NPR footage or a version of the original Outstanding recording. AI likeness, musical gestures and shot timing can vary.
1. Add the people in your set
Choose Solo for one person, or Duet for two. Add one portrait per person and check the browser preview before continuing. For a duet, use portraits with similar camera angles and get the other person’s permission. Do not upload a busy group photo and expect the studio to assign every face a role.
2. Choose the mood and frame
Classic soul uses a warm daytime office. Duet groove emphasizes friendly interactions. After hours uses amber lamps, while Sax spotlight gives the instrumental moment more emphasis. Choose 9:16 for a vertical composition or 16:9 for a landscape composition. All current model tiers request the same 15-second, 720p output and use 20 purchased credits.
3. Sign in, generate, then save
Sign in before generating. If your purchased balance is insufficient, review a one-time pack and its checkout total first. Payment does not automatically start a video: return to the studio, check the portraits and settings, and choose Generate. Follow the existing task rather than submitting repeated attempts. When ready, play and download the result within 24 hours. Confirmed failed tasks return their reserved credits once; uncertain submissions remain reserved while their status is checked.
Sources and editorial notes
Studio instructions describe the current Tiny Desk AI workflow. Creative briefs are optional planning examples; the studio generates with built-in scene and style scripts. Third-party guides link to their official documentation and do not imply an integration or a quality ranking. Read our editorial approach