Alibaba has launched Wan2.1-VACE, an open-source AI model aimed at transforming video creation and editing. Positioned as the first open-source solution in the industry for various video tasks, it promises to streamline the process by integrating multiple tools into one platform. VACE can generate videos from text, images, or other video snippets and features advanced editing capabilities, such as selective modification of video areas without affecting the background. The model allows users to animate still images, control character poses, and expand video dimensions while adding relevant content.
Key technological components include the Video Condition Unit (VCU) for processing multimodal inputs and a Context Adapter structure for better understanding of time and space in videos. Alibaba envisions VACE being useful for social media clips, marketing content, and educational videos. By making this powerful AI tool open-source, Alibaba aims to democratize access, enabling smaller businesses and individual creators to produce high-quality visual content affordably. Both a 14-billion parameter model and a 1.3-billion parameter version are available for free on platforms like Hugging Face and GitHub.
