Wan 2.2 — Image-to-Video (I2V)

Wan 2.2 is a next-gen I2V model built on a Mixture-of-Experts denoising architecture. It turns a single still image into a smooth, cinematic short video with strong prompt adherence and stable motion.

Why it looks great

• Film-grade control: Understands lighting, color, composition, and camera language for cohesive scenes.
• Stable large motion: Handles fast subject/camera movement with fewer jitters or tears.
• Accurate semantics: Follows detailed prompts in complex, multi-object scenes.
• Pure I2V workflow: No start/end keyframes required—one reference image is enough.

Inputs & Parameters

• image (required): Reference image to lock identity, layout, and style.
• prompt (required): Scene mood, motion, and camera cues (e.g., “slow dolly-in, warm rim light”).
• negative_prompt (optional): Things to avoid (e.g., “text, watermark, distortion”).
• size: 832×480 or 1280×720.
• duration: 5 s or 8 s.
• seed: Integer; fixed for reproducibility, −1 for random.
• last_image (optional): The last frame of the video.

How to Use

1. Upload the image.
2. Add a concise prompt (subject + environment + motion + lighting).
3. Choose size (480p/720p) and duration (5 s/8 s).
4. (Optional) Set negative_prompt and seed.
5. Run and download.

Pricing