Wan2.2-I2V-A14B

Wan2.2-I2V-A14B (Wan-AI)

Image-to-video generation focused on turning a single frame into smooth, coherent motion.

• Frame-first design. Takes a still image as input and expands it into a short video, preserving subject identity, style, and composition across frames.

• Large diffusion model. Built as a ~14B-parameter diffusion system, trading heavier compute for better temporal stability and visual detail.

• Consistent motion. Produces more stable motion and fewer flicker artifacts than lightweight image-to-video approaches, especially for character-centric scenes.

• Diffusers-ready. Packaged for the Diffusers ecosystem, making it easier to integrate, test, and swap into existing pipelines.

• Research-grade release. Intended for experimentation and evaluation rather than lightweight or real-time deployments.

Why pick it for Norman AI?

Wan2.2-I2V-A14B is a good fit when visual consistency matters more than speed. It’s useful for testing image-to-video quality, character motion, and scene continuity in higher-end video generation workflows.

response = await norman.invoke(
    {
        "model_name": "wan2.2-i2v-a14b",
        "inputs": [
            {
                "display_title": "Prompt",
                "data": "A cat playing with a ball on mars"
            }
        ]
    }
)