Requires Premium.
Two paths to video
Image-to-Video (recommended) Generate an image, then animate it. This gives you control over the first frame before committing to a video generation. Maximum duration is five seconds. Text-to-Video Write a prompt and generate a clip directly, without a first frame. Faster to start, but less control over the exact composition and character details.The generation loop
Generate a first frame
Use the image generator with the Realism model. Compose the shot carefully — camera angle, character position, and lighting in the first frame carry through to the video. See Workflow.
Open a video session
Start a new Session in the video generator. Sessions keep your generations grouped so you can manage multiple concepts separately. See Sessions.
Select your path
Choose Image-to-Video to animate your first frame, or Text-to-Video to generate from a prompt directly.
Write a motion prompt
Describe the movement and action you want, not just the scene. See Video Prompting.
What affects video output
| Setting | What it does | Where to learn more |
|---|---|---|
| First frame | Sets the starting composition and character | Workflow |
| Motion prompt | Describes movement, action, and camera | Video Prompting |
| Resolution | Controls output dimensions | Image-to-Video |
| Session | Groups related generations | Sessions |
| Enhanced Video | Upscales and sharpens the output | Enhanced Video |
Token cost
Video generation uses tokens. Enhanced Video uses additional tokens. See /premium for current rates.Workflow
Step-by-step: from first frame to finished video.
Sessions
Keep your video generations organized.
Text-to-Video
Generate a clip directly from a prompt.
Image-to-Video
Animate an existing image.