Alibaba introduces a new model for producing cinematic videos

The new model, called Wan2.2-S2V, can produce high-quality video from a single image or audio. It allows you to create videos with a variety of framing options, including model portrait, bust, and full perspective, and features versatile character animation.

Alibaba also states that the model can dynamically generate character movements and environmental factors based on quick instructions. Created videos can be in 480p or 720p resolution.

Wan2.2-S2V can voice-activate global text-driven gesture control, enabling more natural-looking characters. Alibaba noted that another key breakthrough in the model is its innovative frame processing technique. By compressing frames of arbitrary length into a single, compact representation, the model significantly reduces processing power requirements.

On the other hand, the company does not specify how long videos can be produced.