Have you ever been bothered by the complexity of video production and the difficulty of post-editing? Now, Alibaba Tongyi Wan team has brought their latest masterpiece - VACE, an all-round model that integrates video generation and editing, which will completely change your video creation experience.
One of the core highlights of VACE is its powerful on-conditional video generation capability. Just describe the scene in your mind through text, and VACE can quickly transform it into a vivid video. It's like having a professional filming team that is on standby at any time, which can efficiently realize your creative concepts, greatly improving the convenience and efficiency of video production.
VACE not only supports a variety of flexible editing methods, but also provides rich functions to meet different creative needs. For example, with the Move-Anything function, you can precisely control the movement trajectory of any object in the video, whether it is the elegant arc of a bird or the designated route of a car, VACE can easily achieve it. In addition, the Swap-Anything function allows you to replace the subject in the video as you like. Whether it is turning the cheetah in the documentary into a panda or instantly changing the protagonist of the commercial film, VACE can help you achieve the "Great Movement of the World".
VACE also has the Reference-Anything function, which can learn and apply the style of the reference object you provide to the video subject. Whether it is giving the real-life videos anime texture or allowing ordinary people to instantly become the protagonists of artistic paintings, VACE can easily deal with it. The Expand-Anything function can intelligently expand the content of the video screen, solve the problem of limited shooting angles, and make the field of vision wider. The Animate-Anything function can make static images move, and whether it is old photos or artistic paintings, it can bring new vitality under the magic of VACE.
It is worth mentioning that VACE has also shown outstanding strength in video re-rendering. It supports various modes such as content retention, structure retention, subject retention, posture retention and motion retention. Even old videos with blurry pictures and lost details can be reproduced under the repair of VACE, as if time goes back.
The VACE model of Alibaba Tongyi Wan team, with its powerful functions in video generation and editing, will undoubtedly greatly improve the efficiency and fun of video creation. It is like an omnipotent video magician, able to satisfy users' all kinds of imaginations.
Although the model has not been officially launched yet, it will be open source. The arrival of VACE will bring a profound change to the field of video content creation.
Paper link: https://arxiv.org/pdf/2503.07598
Open source code: https://ali-vilab.github.io/VACE-Page/