Kunlun Wanzhi Group has announced the launch of its latest technological achievement, SkyReels-A3 model, which is an audio-driven digital human creation tool based on the DiT (Diffusion Transformer) video diffusion model. The release of SkyReels-A3 marks a major advancement in the field of digital content creation, as it enables full-modal audio-driven digital human creation for any duration, offering users a new experience.
The core function of the SkyReels-A3 model is to bring static images or videos to life. By uploading a portrait image and corresponding voice, the person in the image can speak or sing according to the voice content. Additionally, the model supports the creation of new video content. Users only need to provide a portrait image, voice, and text prompt to generate a video that performs according to the specified state. SkyReels-A3 can also "change the lines" of existing videos, automatically matching new mouth shapes, expressions, and performances, maintaining the continuity of the video.
The model has been optimized and improved in terms of text prompt input, naturalness of action interaction, camera movement control, and video output duration. SkyReels-A3 supports single shot video output up to 60 seconds, and multi-shot support for unlimited duration, meeting different creative needs. Kunlun Wanzhi has also made specific optimizations for practical application scenarios such as online live streaming, improving the consistency of video generation and the naturalness and clarity of specific interactive actions.
The launch of SkyReels-A3 not only provides strong technical support for commercial applications such as advertising and live streaming sales, but also offers more possibilities for artistic creations such as music MVs, movie clips, or speech videos. Kunlun Wanzhi introduced a lens control module based on the ControlNet structure, achieving frame-level precise lens control. It predefines eight common lens parameters, and users can choose the corresponding lens according to their needs, with each lens intensity adjustable continuously from 0-100%, generating professional lens effects.
The release of the SkyReels-A3 model indicates that digital content creation will become more efficient and convenient. Through this innovative technology, Kunlun Wanzhi provides low-barrier, low-cost, high-fidelity AI production solutions for fields such as film production, virtual live streaming, game development, and educational content creation. The launch of SkyReels-A3 represents the possibility of sound as imagery, providing unprecedented efficiency and convenience for the creation of personalized and interactive content. Perhaps the next viral video will come from your inspiration.
SkyReels-A3 Project Homepage:
https://skyworkai.github.io/skyreels-a3.github.io/
SkyReels Official Website Address:
https://www.skyreels.ai/home
SkyReels Series Open Source Model Address:
https://huggingface.co/Skywork