AI VideoCreate
AI video generator for footage from text or a first frame
Describe a scene and the AI video generator makes it as a clip of 5 to 15 seconds. Start from words alone or from a picture as the first frame, choose a camera move and a look, and cut the result in the video editor.
What AI Video can do
Text to video
Describe the place, the subject, the light and what happens, at any length from 10 characters. A description longer than the model reads is shortened to fit by the AI, and the studio says so. Each request returns one clip of that scene.
Image to video from a first frame
Choose a picture from your library, or upload one, as the opening frame. The clip begins on that picture and your prompt says what happens next. The picture can be up to 10 MB.
Six camera directions
Leave the camera on Auto or choose Push in, Orbit, Tracking, Aerial or Locked off. The choice is added to your prompt as a camera instruction.
Six looks
Natural leaves the prompt as you wrote it. Cinematic, Editorial, Product, Analog and Surreal each add a written description of that look.
Six frame shapes
Generate in 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16, at 480p or 768p, for 5 to 15 seconds.
Sound that follows the prompt
There is no sound switch. Say what should be heard in the prompt, and the clip's audio follows it.
A seed for a scene you want again
Give a request a seed number and use it again with the same prompt to help reproduce a scene. Exact results can still vary.
Presets and a library of every clip
Save a prompt with its camera, look and settings as a named preset. Every clip is kept with its prompt in the Video library, and Reuse settings loads any of them back into the form.
Straight into the video editor
A finished clip opens as the first clip of a new edit and is saved to your media library. Inside an edit, the Generate panel adds a 5, 10 or 15 second clip without leaving the timeline.
How AI Video works
Describe the scene
Open AI Video and write what the camera sees and what happens. Add a first frame if the clip should start from a picture.
Set the shot
Choose the frame shape and a length from 5 to 15 seconds. Under Creative direction, pick a camera direction and a look.
Generate the clip
Press the round arrow beside the prompt (Generate video). The request is queued, generated and saved to your library, and its card says which. It keeps going if you leave the page.
Cut it or download it
The clip opens in the video editor, ready to trim and join with other clips. To keep the file as it is, press Download on its card.
AI Video formats and limits
Last updated .
| Detail | AI Video |
|---|---|
| What goes in | A prompt of any length from 10 characters (the AI shortens one past the model's 2,000), and optionally one first-frame picture |
| What comes out | One MP4 clip a request, with audio that follows the prompt |
| Length | 5 to 15 seconds, in whole seconds; 5 unless you change it |
| Frame shapes | 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16 |
| Resolution | 480p or 768p; 768p unless you change it |
| First frame | One image from your media library, up to 10 MB |
| Camera | Auto, Push in, Orbit, Tracking, Aerial or Locked off |
| Look | 6: Natural, Cinematic, Editorial, Product, Analog, Surreal |
| Seed | Optional, from 0 to 2,147,483,647; random when left empty |
| Credits | 480p: 20 credits for 5 seconds, +5 for every 5 more; 720p and 768p: 30, +5; 1080p: 50, +10. Given back if it fails or is cancelled |
Who AI Video is for
Editors missing a shot
Fill a gap in a cut with an establishing shot or a cutaway that nobody filmed.
Marketers trying an idea
See a campaign concept move for five seconds before paying for a shoot.
Teams with one good product photo
Use the photo as the first frame, set the camera to Push in and get a moving version of it.
Social video makers
Generate an upright 9:16 clip as the opening shot of a reel, then add captions and music in the editor.
Directors building a mood film
Generate a shot at a time in one look, and join the clips on a timeline to show how a sequence could feel.
AI video generator questions, answered
What does the AI video generator make?
Short clips of generated footage: a scene that was never filmed, made from your description or from a picture used as the first frame. One request gives one clip of 5 to 15 seconds. It does not write a script or assemble a film with scenes and narration; the AI Product Director does that.
How long can a generated video be?
Between 5 and 15 seconds a clip. For something longer, generate several clips and join them in the video editor, where an edit can run for five minutes.
Can it turn an image into a video?
Yes. Pick one picture from your library as the first frame and describe what should happen. The picture is the opening frame only: you cannot set a last frame or give it several pictures.
What resolution are the clips?
480p or 768p. There is no 1080p or 4K setting for generation. An edit that contains the clip can be exported at up to 4K, which enlarges the clip without adding detail.
Does the generated video have sound?
It can. Describe the sound in your prompt and the clip's audio follows it; there is no separate switch. In the video editor you can turn the clip's volume down and put music or a recording over it.
How many credits does a clip use?
It depends on its resolution and length: a 5-second clip at 480p uses 20 credits, at 720p 30 and at 1080p 50, and every further 5 seconds adds 5, 5 or 10. The studio shows the price before you generate. Exporting an edit that uses the clip is free.
Can I cancel a clip once it has started?
No. A request cannot be cancelled after it is sent, and it has been counted by then. The result is saved to your library when it finishes, whether or not the page is open.
What can the AI video generator not do?
It takes no video as input, so it cannot extend, restyle or re-voice footage you already have. Nothing in it keeps a person or an object identical from one clip to the next; a shared seed and look only help.
AI Video
Footage that never existed, on demand.
Free to start, with 200 credits a month. Paid plans from $10 a month.