All guides

AI Band Video from Photos: What You Need and What You Get

Published 1 November 2026

You hand over a photo of a face. A few minutes later there is a short video of a band playing, with that face singing at the front of it. That is the whole idea. The rest of this page is the detail people usually find out after paying.

What an AI band video from photos actually is

It is not your photo moving. The picture is a reference. What comes back is a freshly generated performance built to resemble the people you uploaded, set in one scene that never changes: five musicians packed in behind a desk in an office, playing to a small room. The result is vertical, just over 19 seconds, and it has sound on it.

Because the performance is generated rather than filmed, nothing travels across from your photo except the likeness. The room, the clothes, the instruments and the movement are all invented. People who expect their own shirt to turn up are the ones who end up disappointed.

Which photos hold up

One adult per picture, head and shoulders, facing the camera, lit well enough that both eyes are visible. JPG, PNG or WebP, up to 10 MB each.

• Focus matters more than how good anybody looks in the picture. Slightly soft and the likeness goes with it.
• Even light. Half a face in shadow comes back muddy.
• No sunglasses, no masks, no heavy filters. Those cover the parts that carry a likeness.
• Face the camera. Side profiles and heads turned away give the weakest results.
• One person per picture. Drop in a group shot and the whole picture gets treated as the person for that seat.

A phone photo is fine. A screenshot of a video call usually is not, because it is softer than it looks on your own screen.

One photo still gets you five musicians

There are five places in the band: a lead singer, someone on keys, someone on saxophone, and two backing singers, one on each side. Only the lead singer has to be filled. Any seat you leave empty is played by a fictional portrait the site keeps for the purpose, so a single uploaded photo still produces a full five-piece band rather than one person alone in a room.

The size of the band has no effect on the price. Five faces and one face cost the same.

What the finished video is like

Vertical, shaped for a phone screen, just over 19 seconds, 24 frames a second. There are two sizes. The smaller one is clear enough on a phone and is what most people end up sending. The larger one holds together better on a laptop screen or a television. Nothing we have downloaded has had a watermark on it, and we checked frame by frame.

Two files come back for every video. One carries the performance audio. The other is the same picture with the sound stripped out, which is what you want if the clip is going into an editing app, or onto a social app that lays its own track over the top.

What it costs and how long it takes

Credits, bought once, with no subscription: $12.99 for 10, $29.99 for 30, $49.99 for 60. A video at the smaller size costs 10 credits and the larger size costs 15.

The smaller size came back in roughly four to six minutes across the videos we timed. The larger size takes considerably longer, and we have timed exactly one of those, so plan for a long wait rather than a particular number. You can close the tab. The video waits in your account.

When a video does not come out, the credits land back in your balance by themselves, without anybody having to ask.

When a photo makes a good band video

The clip is short and the performance is the same every time, so what makes one land is the face you pick and the moment you send it.

• A birthday message where the person out front is the person it is for.
• A send-off for someone leaving, played once at the end of a meeting.
• A group chat joke that works because everybody recognises the faces.
• A small gift for someone who never got to be in a band and would have liked to.

What it will not do

All of it is a reworking of one reference performance that never varies, so none of what follows is waiting on a release. It is how the thing is built.

• The five camera angles and the moments the video cuts between them are the same every time. You cannot reorder them, add one, or hold a shot longer.
• The music is fixed. There is no way to put a different song underneath.
• Faces drift. What comes back looks like the person without being a reliable copy of them, and the movement and the mouth shapes are approximate in the same way.
• Children, pets and groups are out. Every portrait has to be one adult on their own.
• The right to use somebody's face and the right to use music are two separate things, and this site hands you neither. If a real person is going into the band, ask them first.