Image-to-Video

Image-to-video uses a still image as the starting frame of a generated clip. It’s great when you’ve already nailed a hero composition and just want motion added.

Setup

1. Drop an Image node with your starting frame (upload, paste, or generate it on-canvas)
2. Add a Video node and connect the Image into its images handle
3. Add a Prompt node describing the motion and connect it to the Video node’s prompt
4. Pick a video model — Veo 3 for camera control, Kling 2.1 for character motion, Seedance 2.0 for multi-shot

Writing the prompt

Describe motion, not the still:

✅ “Slow dolly forward; mist drifts left to right; subject blinks once.”

❌ “A black-and-white film still of a fox sitting on a windowsill in the morning.”

Mention:

• Camera (dolly, pan, tilt, static, handheld)
• Subject motion (gestures, gait, gaze)
• Environmental motion (wind, rain, smoke, water)
• Mood / pacing (slow, frenetic, dreamlike)

Length and aspect ratio

Set on the Video node before generating. Available options vary by model — for example, Seedance 2.0 supports cinematic 21:9 and lengths up to 12s.

Cost & wait time

Video generations cost more credits and take longer than images (typically 30s–3min). The cost is shown on the node before you click Generate.

After generation

The clip plays inline on the node. From there you can:

• Right-click → Schedule to Instagram to queue a post
• Open it in your Library to download or reuse
• Connect the video into another Video node’s images to extend or restyle it

Tips

• For looping clips, prompt for a “subtle, looping motion” and keep the camera static.
• If a model produces unstable motion on faces, switch to Kling 2.1 and re-run.
• Generate the same prompt on two models side by side for an A/B before committing more credits.