Tiny Desk AI: put your whole band at the little desk
Tiny Desk AI turns up to five photos into a 19-second vertical desk concert video with the original song under it. Choose who sings lead, who sits at the front desk, who stands at the back and who plays sax, and the band is rebuilt around them. A standard video is $9.99.
- 5 seats
- 19 seconds
- 9:16 vertical
- Original song
- From $9.99
Cast the five seats
Only the lead is required. A seat left empty reuses the lead's photo.
100 credits = one standard video = from $9.99. Credits for a failed video are returned.
How to make a Tiny Desk AI video
Three steps and a few minutes. You do not need an editing app, a prompt or a free afternoon: the cast is the only thing you choose.
1Choose who sits where
Open the generator and drop a photo on each of the five seats: the lead singer, the front desk, the two back singers and the sax player. The lead is required; any empty seat reuses the lead's photo.
2Pick standard or HD and confirm
Standard is 480p and costs 100 credits, HD is 720p and costs 300. Tick the box to confirm you have the right to use every photo, then press Make my Tiny Desk video. Sign in when asked; your photos stay on the page.
3Download and post
Both pieces render at once and are joined with the original song under them. When the player appears, download the 9:16 MP4 and post it with a note that it is AI-generated.
What is a Tiny Desk AI video?
The format started as a joke about the Tiny Desk Concerts series, where a band plays at a crowded office desk in front of shelves full of clutter. A short clip of a concert of the song “Outstanding” became the template: fans replaced the performers with friends, teams and rivals, and the result was posted as a vertical video. A Tiny Desk AI video is the same idea made on demand. You supply the faces, and the generator rebuilds the five shots around them.
This site makes one video, and makes it carefully. The original clip is 19 seconds long and has five shots, so it is cut at the real shot change into two pieces. Each piece is rendered by the video model together with your photos, and the two results are trimmed back to the exact length of the pieces, joined, and given the original song again. The picture is generated; the timing, the camera cuts and the music are the original's. That is why every video lands on the same beats as the clip you have seen.
Because the model copies the face, hairstyle and clothes of each photo, the choice of photo matters more than anything you could type. The other thing that matters is the cast. The original has five people, so there are five seats. If you want the whole video to be about one person, upload one photo; if you want an entire team, upload five. Either way you see the price first and you pay only for a video that finishes.
Who sits where
The template has five seats. Each one is a different person in the original, and each gets its own photo. Seats 1 to 4 appear in both halves of the video; the sax player in seat 5 only joins in the second half.
- 1
Lead singer
Opens the video, then sings and laughs from the right of the frame.
- 2
Front desk
Right at the camera, sings and plays the keys.
- 3
Back left
Small in the back, left of the frame.
- 4
Back right
Small in the back, right of the frame, fists up.
- 5
Sax
Plays the saxophone in the back, left of the frame.
The five shots of the Outstanding desk concert
The original is a five-shot sequence, which is also how it is described in most of the videos you have seen. Knowing the shots tells you which photo to spend your best picture on: the lead is on screen for most of the video, so it should be the clearest face you have.
| Time | Shot | Seats on screen | What happens |
|---|---|---|---|
| 0:00 – 0:01.4 | Lead close-up | Lead singer | The lead singer looks off to the left with lips pressed together, then leans toward the camera and parts their lips. One seat, one face. |
| 0:01.4 – 0:06.7 | The desk joins in | Front desk, back left, back right | The camera pulls back to the desk. The front seat sings and claps close to the lens while the two back seats sing and pump their fists. |
| 0:06.7 – 0:13.0 | The laugh | Lead singer | A long close-up of the lead as a grin turns into a hearty laugh, a raised hand and a loose fist. This shot decides how good your video looks, so pick the best photo for the lead. |
| 0:13.0 – 0:16.4 | Sax solo | Sax, front desk | The saxophone player blows into the horn at the back while the front seat nods along and opens their mouth. |
| 0:16.4 – 0:19.1 | Band finale | Lead, front desk, back left, back right | All four singers on screen. The lead points up between the back seats, the front seat sings into the keys, and the lead finishes with a fist in the air. |
Photos that work, and photos that do not
Use
- ✓One adult, face fully visible and front-facing
- ✓Even daylight or soft front light
- ✓Head and shoulders, with a collar or shirt in frame
- ✓Neutral or lightly smiling expression
Skip
- ✕Sunglasses, masks or anything over the eyes
- ✕Group photos: the model has to guess who you mean
- ✕Heavy filters, beauty effects or tiny, blurry faces
- ✕Minors, celebrities or anyone who has not agreed
Getting a better result from your photos
A good photo is a clear, front-facing picture of one adult, in daylight or evenly lit, with the whole face visible. A head and shoulders crop is ideal, because the model copies clothing too, and a visible collar or shirt gives it something to work with. Neutral or slightly smiling expressions work best: a mid-sentence face or a wide grin can come out strange when the video asks the person to do something else.
Photos that fail are predictable. Sunglasses and masks hide what the model needs to copy. Group photos make it guess which person you mean. Heavy filters, beauty effects and very low resolution pull the face toward an average. A face that fills less than a fifth of the picture is too small. If a first video comes out wrong, change the photo, not the settings: it is almost always the photo.
Lighting and angle count too. A face that is lit from the front and slightly above looks natural when the video turns it toward the camera; a face that is lit hard from one side can look odd when it turns. If you are casting a group, take the five photos in the same kind of light, so the band looks like a band. Try the standard version first, change anything you do not like, and spend the HD price on the cast you finally keep.
Want to know who is behind the site? Read the About page. Not sure how many photos to use? See the Outstanding Tiny Desk AI guide for the cast ideas people use, or go straight to pricing.
Who makes a Tiny Desk AI video, and why
Friends and family
The classic use. Put your group chat on the desk, give your sister the lead, and send the clip back to the chat before the original joke wears off.
Teams and workplaces
A team clip for a farewell, a launch or an all-hands is more fun when the people in it are the people in the room. Ask everyone first and use their own photos.
Rivals and fan groups
Sports fans use the format to put rival coaches or their own favorites in one band. Only use people who are fine with it, and label the result AI-generated.
Solo "I am the whole band"
Upload only the lead photo. Every empty seat reuses it, so you get the same person on every instrument, which is a joke of its own.
Tiny Desk AI compared with a single-photo generator
Most Outstanding generators take one lead photo. That is fine for a solo joke, but the format works because it is a band. Here is what is different here.
| Tiny Desk AI | Typical single-photo generator | |
|---|---|---|
| Cast | Five seats, each with its own photo | One lead photo, sometimes a group photo |
| Price shown before sign-in | Yes, on every page | Often only after login |
| Failed video | Credits returned automatically | Terms vary; often a support request |
| Output | 19 s, 9:16 MP4, original song | 10–15 s, 9:16 MP4 |
| Quality choice | Standard 480p or HD 720p | Usually one |
| Subscription | None; credits never expire | Often packs plus monthly plans |
Prices you can see before you sign in
Credits are bought once and never expire. A standard video (480p) is 100 credits, an HD video (720p) is 300. If a video fails, the credits are returned automatically.
Single
$9.99
1 standard video
Trio
$24.99
3 standard videos, or 1 HD video
Full band
$59.99
8 standard videos, or 2 HD and 2 standard
Use it responsibly
- Only upload photos of adults who have agreed to appear. We do not allow minors, celebrities or public figures.
- Do not use a video to deceive, harass or impersonate anyone. Label what you post as AI-generated.
- Your photos are used only to make your video and are not used to train models.