Add up to five photos — three singers, a lead and a saxophone — and get a 15-second vertical performance video with original music, filmed on a warm little record-store stage.
No prompt writing · No watermark · Failed renders refunded automatically
Record-store studio · 15s · 9:16
Two photos in · music included
Five photos in, one finished performance out. The stage, the camera moves and the music are already produced — you only supply the band.
Your tiny desk session
Add five photos to start
Three singers wide, one close-up solo, one saxophone finish.
Example
The five stills below are the only thing the user uploaded. The room, the framing, the camera moves and the soundtrack were already produced.
The five photos in
Straight from a phone, nothing edited
Five stills go in; the stage, the camera moves and the soundtrack are already built. Previewing a frame of your own set is not charged — the finished 15-second file is what you unlock.
The five photos in
Straight from a phone, nothing edited
Five stills go in; the stage, the camera moves and the soundtrack are already built. Previewing a frame of your own set is not charged — the finished 15-second file is what you unlock.
Everything up front: what you upload, what you get, how long it takes, what it costs.
Input
5 photos
Three group singers, a lead and a saxophone player. JPG, PNG or WebP up to 8 MB, one clear face each.
Output
15s MP4
9:16 vertical, original AI soundtrack, no watermark, saved in History and downloadable.
Time
3–6 min
The render runs on the server. You can close the tab and collect it from History.
Price
Preview, then pay
Generating a watermarked preview costs nothing. Unlocking the finished video is a one-time purchase: 50 credits = 1 video, sold as $5.99 (50 credits) / $9.99 (100) / $14.99 (200), valid 30 days.
The script is already written. You are only casting it.
Left singer, center, right, lead and saxophone. Clear, front-on faces work best — reusing the same photo in several slots is fine.
The stage, the lighting, the camera moves and the soundtrack are already produced. Nothing to configure.
A 15-second vertical MP4 with music, delivered in 3–6 minutes. No watermark, yours to post.
Three camera setups, cut to one continuous track.
00:00
Three singers shoulder to shoulder in front of the record shelves, mouths open on the downbeat.
00:05
The camera drifts in until the lead singer fills the frame, background falling into soft bokeh.
00:11
Cut to the closing solo: the sax player leans into the mic stand, the band still going behind.
00:15
The clip lands on the last note — no fade, no watermark, ready to post.
Definition
A Tiny Desk AI video is a 15-second vertical performance clip — three singers, a solo close-up and a saxophone finish — generated from five of your own photos. The set is a warm record-store stage with wooden shelves, a red analogue synth and a microphone stand; the music is original, so nothing you post touches a copyright claim.
The script, camera moves, lighting and soundtrack are produced once and reused for every video, so there is no prompt to write and no timeline to edit. You cast five roles, press one button, and receive a finished vertical MP4.
Left, center and right singers, a lead, and a saxophone player.
The blocking, lighting and grade are pre-tuned and locked.
An AI-composed soul-funk groove — not a licensed or covered song.
720p MP4 sized for TikTok, Reels and Shorts.
Every clip is clean and ready to post as your own.
Credits are bought once. No subscription, no auto-renewal.
Most sessions finish in a few minutes, even if you close the tab.
If a render cannot finish, credits return to your balance.
Under the hood
Most photo-to-video tools re-imagine the whole scene on every run, so two videos from the same upload never look alike. Here the set is generated once, reviewed, and then frozen as a fixed background plate for every job.
Each job only does one thing: place your five subjects onto that plate, then animate it. That is why the room, the props and the light stay identical between your video and the example above.
Three plates — wide, close-up, saxophone angle — from the same room.
Each upload is used as the reference for its own role.
The same 15-second groove runs under all three cuts.
Warm amber practicals and light film grain, baked into the plates.
Anyone with five photos and a group chat.
Cast the whole group chat as a band that never rehearsed.
Five dogs, five roles. The most popular version by far.
A leaving gift, an offsite opener, or a launch announcement.
A repeatable format you can post weekly without re-editing.
Previews and unlocked videos from this browser. Renders keep going if you close the page.
Not signed in
Sign in to buy credits and unlock videos
No sessions yet. Add five photos to start.
One finished video costs 50 credits. Generating a preview is not charged. Credits are a one-time purchase — not a subscription, never auto-renewed — and stay valid for 30 days.
Credit packs
50 credits = 1 finished video. One-time payment, valid 30 days.
50 credits = 1 finished video · 50 credits each · one-time payment
100 credits = 2 finished videos · 50 credits each · one-time payment
200 credits = 4 finished videos · 50 credits each · one-time payment
50 credits unlock one finished video. Credits are bought once, never auto-renew and expire 30 days after purchase. If a render fails or is rejected by content review, the credits return to your balance. Purchases are subject to the Terms of Service and Content Standards. · Balance: 0 credits · 50 credits unlock one video.
What you upload, what you get, and what the render actually looks like.
Five photos, one per role: left singer, center singer, right singer, lead singer and saxophone player. JPG, PNG or WebP up to 8 MB. A clear, front-facing face in each one works best.
You can reuse the same photo in several slots — the same person can sing three parts, or one photo can fill the whole band. Every slot still needs a file before the render starts.
You don't get the original Outstanding recording from us — it is never used, sampled or covered. Every render is built from your five photos and carries a 15-second soul-funk groove composed by an AI music model for this template. If you want that specific song in the video, download the watermark-free MP4 and lay your own audio under it in an editor; the file you get has no rights attached to anyone else's track.
Usually 3–6 minutes: three keyframes, three video clips, then the final cut with the soundtrack. You can close the page — the job keeps running and appears in History.
Yes. A continuous 15-second track runs under all three cuts, so the edits land on the music rather than on arbitrary frames.
Credits go back to your balance automatically. Nothing is lost and there is no ticket to open.
One finished video costs 50 credits. Credits are sold once — $5.99 for 50, $9.99 for 100, $14.99 for 200 — and stay valid for 30 days. Generating a preview uses no credits, and there is no subscription or auto-renewal.
Sexual or NSFW content, violence and gore, hate and harassment, anything involving a minor, deepfakes or impersonation of a real person, and material that infringes copyright or trademark. Every uploaded photo and every generated frame is screened before rendering; blocked renders are not charged. The full list is on the Content standards page.
Black Forest Labs FLUX 3 for the still frames and for image-to-video, ACE-Step for the soundtrack. The model and resolution are recorded in each job's metadata.
Yes, but disclose it. TikTok, Instagram and YouTube all require synthetic-media labels, and the platform toggle takes two seconds — the video still performs the same.
They are stored under random private paths for the duration of the render and used only to build your video. They never appear in a public gallery, and no example on this site uses a customer's photos.
Five photos, one button, one finished fifteen-second performance with its own soundtrack.