Create videos from text or from an existing image directly on the canvas workspace.
You can generate videos directly on the canvas in two ways: from a text prompt (text-to-video) or from an existing image on the canvas (image-to-video). Both give you control over model selection, duration, aspect ratio, and audio.
Generating videos directly on the canvas lets you pick the exact model, duration, and settings. For simple generation tasks, this uses fewer credits than asking the agent in the chat.