Create videos

Agent can create a video using only text, animate an image, or use two images to guide how a video begins and ends.

Create a video using only text

Describe the subject, setting, action, and camera movement. Agent will create the video without a start image.

For example:

Create a 5-second cinematic video of a silver handbag on a rotating pedestal in a dark showroom. Use a slow camera push-in, soft moving reflections, and one continuous shot.

Animate an image

Attach or reference the image you want to animate, then describe the motion.

The image establishes the subject, product, setting, and starting composition. Focus your request on how the subject and camera should move, and what should remain still.

For example:

Turn this image into a 5-second video. The model slowly turns to show the back of the jacket while the camera moves in a gentle arc.

Change the image before animating it

If the start image also needs a change, include the new visual result and the motion in the same request. Agent will prepare the image first, then create the video.

For example:

Place this handbag on a marble pedestal in a dark showroom, then create a 5-second video with a slow camera push-in and moving reflections.

Use a start image and an end image

Use two images when you want more control over how the video begins and ends. The start image sets how the video begins, while the end image guides how it ends.

Attach or reference both images and identify their roles in your request.

For example:

Begin with the first image and end with the second image. Over 5 seconds, the model turns naturally while the camera moves from the front view to the final side view.

Choose images that can be connected through a believable camera movement or subject action. Keep the same main product or subject visible across both images.

Guide the motion

Keep the request focused on one continuous shot. Describe:

• How the subject or product moves.
• Whether the camera stays still, pushes in, pans, or moves around the subject.
• Which parts of the scene should remain still.
• The final position or composition when it matters.

Use a start image and an end image when you need more control over how the video ends.

Add audio

Videos are silent unless you ask for sound. To add audio, describe what should be heard in the same request: speech, music, ambience, or the sounds of the action.

For example:

Create a 5-second video of the model walking toward the camera through the concrete hall. Add the sound of her footsteps on the polished floor.

• Speech: Write the exact words in quotation marks. Speech works best in English.
• General sound: If you ask for sound without describing it, Agent adds subtle ambience and the natural sounds of the action.
• Silent video: Ask for a silent video at any time to turn audio off again.

A video with audio costs twice the credits of the same video without audio. See video credit costs.

Choose the duration and resolution

Agent supports the following video options, subject to your plan:

When Generation mode is selected, Fast uses 480p, Balanced uses 720p, and Quality uses 1080p. Auto lets Agent choose. See Generation settings.

Limitations

• Existing videos cannot be used as inputs.
• Generated videos cannot be used as image references in later requests.
• Videos cannot be transparent. Transparent areas in a start image or an end image are filled with solid black or white.

What's Next?

• Learn how to reference a start image and an end image.
• Create or edit a start image before adding motion.
• Review generation settings for video quality and resolution.