How to Make a Tiny Desk AI Video

You will get a short vertical band clip built from the photos you upload. The finished file is 15 seconds long, 720p, 9:16, with sound baked in. You can play it in your library, download the MP4, or share it from the site.

On Tiny Desk Band that means an Outstanding Tiny Desk–style desk-stage performance: either one lead singer (Solo Cast) or a three-person lineup (Full Band). Nothing here is a project file for CapCut or Premiere. When you want to make your own band video, you stay inside the studio flow—photos in, MP4 out.

Before you upload anything, choose a mode. That choice locks how many photos you need and who appears on screen.

Solo Cast uses one photo. That face becomes the lead singer. Use this when you only have a clear shot of yourself (or one friend), or when you want a single-voice clip for a story, a birthday card, or a quick test of the look.

Full Band is the default and the main pitch of the product. It takes up to three photos: lead vocal, keys, and saxophone. Only the lead is required, and each extra photo swaps one more musician, so the frame reads as a small band rather than a lone singer. Use Full Band when you can gather three adult faces who agreed to appear—roommates, a friend group, coworkers, or a family of adults.

You can switch modes before you generate. Packs are not tied to a mode: a video credit from any pack can run Solo Cast or Full Band. If you are unsure, start Full Band only when you already have three usable photos lined up; otherwise start Solo Cast so you are not stuck hunting for two more faces mid-checkout.

Open the Outstanding Tiny Desk AI studio, pick your mode, then upload into the role slots you see on screen.

Photo rules that actually matter on a phone screen:

• Face forward. A clear front view of the face works better than a dramatic profile. Soft three-quarter angles can work; hard side profiles often warp when the model tries to sing.
• Even light. Window light or soft indoor light beats harsh overhead bulbs and neon. Avoid heavy shadows across one eye or a blown-out forehead.
• Enough resolution. Cropped thumbnails and heavily compressed chat screenshots look soft once they fill the frame. Prefer the original camera file when you can.
• One person per photo. Do not squeeze a group shot into a single role. The slot expects one face. Crowded frames confuse who should sing.

For Full Band, drop each photo into the matching role: Lead vocal in the center slot, Keys for the keyboard player, Saxophone for the sax player. Confirm the labels before you pay. If two friends look similar, assign the strongest front-facing shot to Lead so the eye has an anchor.

Age and consent are hard requirements: you must be 18 or older, and every person in the photos must be an adult who agreed to appear. Do not upload children or celebrity likenesses you do not have rights to use. Uploaded photos are removed within 24 hours; finished videos stay available for 30 days.

When the roles look right, buy a one-time pack. There is no subscription and no free trial quota.

Credit packs (USD, one-time):

• 10 credits — $6.99
• 50 credits — $29
• 250 credits — $99
• 900 credits — $199
• 3,000 credits — $499

A standard 720p clip with just the lead singer costs 10 credits, and each extra band member adds 4. The studio shows the exact price before you generate. Checkout is a standard payment screen, and credits do not expire.

Before you generate, skim the role list one more time. Swapping a weak Lead photo afterwards means generating again. If you plan a friend-group batch, the 50-credit pack covers about five standard clips, which usually matches that session better than buying 10 credits twice.

Then hit generate. Watch the studio or library for status. When the render finishes, open the clip, play it with sound, download the MP4, or share it. If something looks off (wrong role, soft face), fix the photo or the role map and spend another credit rather than forcing a bad upload through again. Credits sit on the account until you use them; finished files remain downloadable for 30 days.

Most people should default to Full Band when they have three solid photos. Solo Cast is the right call when you only have one face or you want a fast personal clip.

Full Band is how the product is meant to feel: a desk-stage lineup, not a karaoke close-up. Solo Cast still uses the same 15-second, 720p, 9:16, with-sound delivery—you simply cast one singer instead of three. The price follows the number of people you cast: 10 credits for the lead, plus 4 for each extra band member.

If your group can only muster two clear faces today, do not invent an extra from blurry group crops. Run Solo Cast for the strongest face, or wait until you have three separate adult portraits. When you do have the set, map the strongest front-facing shot to Lead and keep the Keys and Sax photos in a similar brightness range so the lineup feels like one night, not three random years. For role layouts and casting tips after you finish this walkthrough, see the band templates guide.

Three photo mistakes kill most first renders. Fix them before you spend a credit.

1. The photo is too soft or tiny. Chat crops and zoomed screenshots turn into mush when the face fills the vertical frame. Mouth and eye detail disappear, and the singing motion looks smeared. Fix: Use the original camera still, keep the face large in frame, and skip filters that melt skin texture.

2. Several people share one photo. A group selfie shoved into Lead (or any single slot) confuses who should move. You may get a blended face or a singer who never quite matches anyone. Fix: Crop to one person per file, or take three separate shots. Full Band is three files, not one panoramic.

3. A hard side profile. Extreme profiles hide the mouth shape the model needs for singing. Results often look stiff or “sliding” on the jaw. Fix: Reshoot facing the camera, or pick a frame where both eyes and the mouth are visible. Soft three-quarter is usually fine; full profile is not.

If you only remember one rule: one clear, front-lit adult face per role.