XunGuang: Alibaba DAMO Academy's AI Video Creation Platform 132

Xunguang is a one-stop AI video creation platform launched by Alibaba's DAMO Academy. Xunguang made its debut at the 2024 World Artificial Intelligence Conference (WAIC) in Shanghai, attracting widespread attention. Xunguang aims to make the video editing process as simple as operating PPT, enabling even non-professional users to efficiently create high-quality video content.

Xunguang helps users complete steps traditionally scattered across different production processes—such as scriptwriting, storyboard design, and video material editing—on a single platform, assisting users in creating scripts and storyboards.

Xunguang supports rich AI editing on generated and uploaded materials, offering more than 10 functions such as character control, scene control, camera movement control, and object addition/removal/modification. In addition, it focuses on "making video editing as simple as operating PPT" in terms of interaction, and provides video layer-based editing capabilities as well as a layer breakdown function.

Xunguang official website:https://xunguang.damo-vision.com/

Xunguang core features:

• One-stop service: The Xunguang platform integrates all aspects of video production, including scriptwriting, storyboard design, and material editing, allowing creators to complete the entire video creation process on one platform without switching between multiple tools.

• AI editing capabilities: The platform provides refined editing tools that allow users to edit at the semantic level, such as controlling video content through text commands to achieve intelligent control over characters, scenes, camera movements, and more. Users can easily adjust camera angles, remove or modify target objects, and even control precise facial movements.

• Video layer editing: Xunguang's innovation lies in its video layer editing function. Users can easily extract and edit specific objects in a video, such as people or objects, which will come with a transparent background for easy integration into other backgrounds. Even complex details such as flowing hair can be handled with precision. Whether it's AI-generated or live-action material, various video editing functions can be used for modification, with layer breakdown and fusion functions available to customize your video content.

• Style transfer and camera movement control: The platform supports more than 20 style transfers, making video style transformation simple. At the same time, it provides a variety of camera movement effect controls, such as panning, pushing in/pulling out, and orbiting, enhancing the visual effects of the video.

• Script and storyboard assistance: Xunguang not only simplifies post-production editing, but also assists users in pre-production scriptwriting and storyboard creation, greatly improving video creation efficiency and creative space.

• Storyboard creation with one click: From script to storyboard, Xunguang lets AI assist your directing journey. No more relying on imagination—the visuals emerge naturally, bringing great stories to life instantly.

• Customize your story characters: Who can become the lead in your story, delivering outstanding acting? Now Xunguang customizes them exclusively for you—your actors are under your control.

• Free creation of visual materials: Now let Xunguang generate consistent characters and scene images according to your ideas, then use camera movement control and motion editing to create your personal blockbuster.

DAMO Academy Xunguang video creation platform beta application requirements:

• General users: UGC groups represented by ordinary enthusiasts

• Semi-professional users: PUGC groups represented by self-media or small studios

• Professional users: PGC groups that use standardized workflows and deliver team-based commercial output

Xunguang is developed by the DAMO Academy's Vision Technology Laboratory, which is dedicated to the research of multimodal visual signal understanding and generation technologies. Its current key research directions include more accurate image/video/3D content generation, more controllable image/video/3D content editing, more efficient generation frameworks, multimodal retrieval-augmented generation, and unified multimodal understanding-generation frameworks.

Xunguang official website:https://xunguang.damo-vision.com/