Image to Video
Breathe life into a static image — turn any picture into a 5–10 second animation.
Image to Video (i2v) turns a static image into a living clip up to 10 seconds long. Start from any generated task via the Animate button, or open the Image to Video page and upload an image from your device.
The settings that matter
Model:
- v2.7 (High Dynamics) — cinematic: camera movement, motion blur, layered depth. Dramatic results even from simple prompts. - v2.6 (High Resolution) — the reliable choice: clean, stable, consistent, minimal artifacts. - v2.5 (High Flexibility) — expressive: subtle motion, rich facial expressions, realistic physics. Also the only model that supports video LoRAs (dance, live-2D, shake).
Mode: Professional for quality (recommended), Basic for speed and fewer Credits.
Duration: 5 or 10 seconds — start with 5 while testing; longer clips are harder to render cleanly.
Writing a video prompt
The formula: Subject + Motion + Environment.
> "A white-haired girl with cat ears gently adjusts her bangs, tilting her head slightly, in a classroom bathed in afternoon sunlight, dust particles dancing in the golden light."
Subject: a brief anchor description (hair, outfit) so features stay consistent while moving.
Motion: the soul of the clip — use specific, physical verbs ("leans forward to pet the cat"), never vague ones ("moves around").
Environment: tells the model how light, wind, and particles should behave.
Camera (optional): direct it like a film shot — "camera slowly pushes in…". Use pacing words (slowly, smoothly) and don't combine opposing moves. Camera control works best on v2.7 (v2.5 doesn't support it).
Helpful extras
AI Smart Prompt (toggle in the prompt box) enhances your prompt automatically; Action Presets give one-click motions like "Kiss" or "Hug"; negative prompts (blurry, disfigured, grainy) keep clips clean.