# Image → Video (I2V) and Reference-to-Video

1. **Select the image** node on the canvas you want to animate.
2. **Switch the category pill to Video.** The chat uses your selected canvas image as the input frame (image-to-video) along with your prompt; set aspect ratio and duration in the chat controls. Cost scales with duration.
3. **Generate.** The video lands as a video node on the canvas.

For richer, multi-reference video (multiple image / video / audio references in one shot), use the **Framo** agent with a cinematic video skill — load **Cinematic Prompt** (or **Cinematic Prompt — Seedance** for the deep reference stack), attach references in the chat, and it runs reference-to-video against them. See [Generating videos](/docs/product/video) and [The Agent (Skills Chat)](/docs/product/agent).