Wan 3.0

Wan 3.0 is Alibaba’s video model. Give it a prompt and it generates a short video with cinematic motion and camera movement. You can also provide a starting image to animate.

What it’s good at

• Text to video. Turn a description into a video clip.
• Image to video. Provide a first-frame image and the model animates it, guided by your prompt.
• Multiple resolutions. Choose 480p, 720p, or 1080p.
• Long clips. Generate up to 30 seconds.
• Adaptive framing. Let the model pick the best aspect ratio for your prompt, or set one explicitly.

Inputs

• prompt: the video you want.
• image: optional first-frame image to animate, as jpg, png, bmp, or webp. When you provide one, the model animates it.
• resolution: 480p, 720p, or 1080p (the default).
• aspect_ratio: adaptive (the default) lets the model choose, or set 16:9, 9:16, 1:1, 4:3, or 3:4. The model ignores this when you provide an image.
• duration: length of the video in seconds, from 2 to 30 (default 5).
• negative_prompt: what to leave out of the video.
• enable_prompt_expansion: rewrites short prompts into more detailed ones. On by default.
• seed: set a seed to get the same result again.