WAN 2.2 Image-to-Video Model (1080p)
WAN 2.2 is an advanced image-to-video model provided by Cloud's DashScope platform. This model adopts an innovative MoE (Mixture of Experts) architecture for generating high-quality video content from images at full HD 1080p resolution.
Why it looks great
- Cinematic control: handles lighting, color, and composition with a film-style aesthetic.
- Stable, natural motion: restores nuanced subject movement and camera dynamics.
- Semantic fidelity: follows complex editing prompts while preserving scene coherence.
- Clean edits: supports negative prompts and prompt optimization to reduce artifacts.
Limits and Performance
- Input: one image + text prompt
- Output resolution: 1080p (Full HD)
- Clip length per job: 5 seconds
Pricing
Per-job billing with a 5-second clip length.
| Duration | Resolution | Cost per job |
|---|
| 5 s | 1080p | $0.80 |
How to Use
- Upload Image – drag & drop or paste a URL.
- Write Prompt – describe motion, mood, camera, lighting, etc.
- (Optional) Negative Prompt – exclude unwanted objects/styles.
- (Optional) Seed – fix for reproducibility (
-1 = random).
- Run – generate, preview, and download.
Notes
- Please check that your uploaded image URL is valid and accessible.
- If the URL is not working, consider uploading the image locally.
- Recommended image formats: JPG / PNG.